Datasets
Models
Downloads: 1
Diff of
/ipynb/embedding 3d visualisation.ipynb
[000000]
..
[98867e]
Switch to side-by-side view
--- a +++ b/ipynb/embedding 3d visualisation.ipynb @@ -0,0 +1,2402 @@ +{ + "cells": [ + { + "cell_type": "markdown", + "metadata": {}, + "source": [ + "## Exploration of 100d space of genome vectors\n", + "\n", + "Genome vectors created by the Dna2VecDataBunch exhibit piculiar patterns. This notebook is dedicated to exploratoin \n", + "of the bacterial genome space using dimensionality reduction techniques" + ] + }, + { + "cell_type": "code", + "execution_count": 2, + "metadata": {}, + "outputs": [], + "source": [ + "%load_ext autoreload\n", + "%autoreload 2" + ] + }, + { + "cell_type": "code", + "execution_count": 3, + "metadata": {}, + "outputs": [], + "source": [ + "import sys\n", + "sys.path.append(\"../mylib/\")\n", + "\n", + "from genomic import sequence\n", + "from genomic.sequence import regex_filter, count_filter\n", + "from functools import partial\n", + "import pandas as pd\n", + "import numpy as np\n", + "from sklearn.decomposition import PCA\n", + "from sklearn import manifold,neighbors\n", + "from scipy.cluster.hierarchy import dendrogram, linkage \n", + "from matplotlib import pyplot as plt\n", + "import seaborn as sns; sns.set(color_codes=True)\n", + "import plotly.plotly as py\n", + "import plotly.graph_objs as go" + ] + }, + { + "cell_type": "markdown", + "metadata": {}, + "source": [ + "### Load Data" + ] + }, + { + "cell_type": "code", + "execution_count": 4, + "metadata": {}, + "outputs": [ + { + "name": "stdout", + "output_type": "stream", + "text": [ + "CPU times: user 50.9 s, sys: 1.61 s, total: 52.5 s\n", + "Wall time: 55.3 s\n" + ] + } + ], + "source": [ + "filters=[partial(regex_filter, rx=\"Streptomyces|Bacillus|Vibrio|Streptococcus|Rhizobium|Staphylococcus\"),partial(regex_filter, rx=\"plasmid\", keep=False),\n", + " partial(count_filter, max_count=599)]\n", + "data = sequence.Dna2VecList.from_folder(\"/data/genomes/GenSeq_fastas/train\",filters=filters,agg=partial(np.mean, axis=0),n_cpus=7)\n", + "processors = [\n", + " sequence.GSFileProcessor(),\n", + " sequence.GSTokenizeProcessor(tokenizer=sequence.GSTokenizer(ngram=8, skip=0, n_cpus=7)),\n", + " sequence.Dna2VecProcessor()]\n", + "%time for p in processors: p.process(data)" + ] + }, + { + "cell_type": "code", + "execution_count": 4, + "metadata": {}, + "outputs": [ + { + "data": { + "text/plain": [ + "3169" + ] + }, + "execution_count": 4, + "metadata": {}, + "output_type": "execute_result" + } + ], + "source": [ + "len(data.items)" + ] + }, + { + "cell_type": "markdown", + "metadata": {}, + "source": [ + "### Genome vectors" + ] + }, + { + "cell_type": "code", + "execution_count": 5, + "metadata": {}, + "outputs": [], + "source": [ + "def log_scale(X):\n", + " x=np.asarray(X);e=1e-6\n", + " return np.log10(x+np.abs(x.min())+e) \n", + "\n", + "\n", + "x=np.asarray(data.items)\n", + "bad_fastas = np.where(np.mean(x,axis=1) == 0.)[0]\n", + "X = np.delete(x, bad_fastas,0)\n", + "labelList=[\" \".join(i.split()[1:3]) for i in data.descriptions]\n", + "labelList=np.delete(np.asarray(labelList), bad_fastas)\n", + "vocab=list(np.unique(labelList))\n", + "y=[vocab.index(x) for x in labelList]" + ] + }, + { + "cell_type": "code", + "execution_count": null, + "metadata": {}, + "outputs": [], + "source": [] + }, + { + "cell_type": "markdown", + "metadata": {}, + "source": [ + "## Correlation Distance in log-scaled space" + ] + }, + { + "cell_type": "markdown", + "metadata": {}, + "source": [ + "### tSNE" + ] + }, + { + "cell_type": "code", + "execution_count": null, + "metadata": {}, + "outputs": [], + "source": [ + "tsne = manifold.TSNE(n_components=3, init='pca', perplexity=10, metric=\"correlation\",random_state=0)\n", + "%time X3 = tsne.fit_transform(log_scale(X))" + ] + }, + { + "cell_type": "code", + "execution_count": 12, + "metadata": {}, + "outputs": [], + "source": [ + "genus = [i.split()[0] for i in labelList]\n", + "genus_vocab=list(np.unique(genus))\n", + "y=[genus_vocab.index(x) for x in genus]\n", + "genus_vocab\n", + "\n", + "X3_df = pd.DataFrame(data=X3, columns=[\"pc1\",'pc2','pc3'], index=labelList)\n", + "\n", + "X3_df[\"genus\"]=genus\n", + "X3_df[\"y\"]=y\n", + "\n", + "genus_df=X3_df.groupby(\"genus\").agg({\"pc1\": list, \"pc2\":list,\"pc3\":list,\"y\":np.mean})" + ] + }, + { + "cell_type": "markdown", + "metadata": {}, + "source": [ + "### Correlation Distance visualisation" + ] + }, + { + "cell_type": "code", + "execution_count": 16, + "metadata": {}, + "outputs": [ + { + "name": "stderr", + "output_type": "stream", + "text": [ + "/home/serge/anaconda3/envs/bio/lib/python3.6/site-packages/IPython/core/display.py:689: UserWarning:\n", + "\n", + "Consider using IPython.display.IFrame instead\n", + "\n" + ] + }, + { + "data": { + "text/html": [ + "<iframe id=\"igraph\" scrolling=\"no\" style=\"border:none;\" seamless=\"seamless\" src=\"https://plot.ly/~sergeman/6.embed\" height=\"1000px\" width=\"1000px\"></iframe>" + ], + "text/plain": [ + "<plotly.tools.PlotlyDisplay object>" + ] + }, + "execution_count": 16, + "metadata": {}, + "output_type": "execute_result" + } + ], + "source": [ + "data=[]\n", + "for g in genus_df.index:\n", + " trace = go.Scatter3d(\n", + " name = str(g),\n", + " x=genus_df.loc[g,\"pc1\"],\n", + " y=genus_df.loc[g,\"pc2\"],\n", + " z=genus_df.loc[g,\"pc3\"],\n", + " mode='markers',\n", + " marker=dict(\n", + " size=8,\n", + " color=genus_df.loc[g,\"y\"], # set color to an array/list of desired values\n", + " colorscale='Jet', # choose a colorscale\n", + " opacity=0.5)\n", + " )\n", + "\n", + " data.append(trace)\n", + " \n", + "\n", + "layout = go.Layout(\n", + " width=1000,\n", + " height=1000,\n", + " margin=dict(\n", + " l=0,\n", + " r=0,\n", + " b=0,\n", + " t=0\n", + " )\n", + ")\n", + "fig = go.Figure(data=data, layout=layout)\n", + "py.iplot(fig, filename='correlation distance metric by genus')" + ] + }, + { + "cell_type": "markdown", + "metadata": {}, + "source": [ + "## Eucleadian Distance in log-scaled space" + ] + }, + { + "cell_type": "markdown", + "metadata": {}, + "source": [ + "### tSNE" + ] + }, + { + "cell_type": "code", + "execution_count": 20, + "metadata": {}, + "outputs": [ + { + "name": "stdout", + "output_type": "stream", + "text": [ + "CPU times: user 1min 13s, sys: 308 ms, total: 1min 14s\n", + "Wall time: 1min 13s\n" + ] + } + ], + "source": [ + "tsne = manifold.TSNE(n_components=3, init='pca', perplexity=30,random_state=0)\n", + "%time X3 = tsne.fit_transform(log_scale(X))" + ] + }, + { + "cell_type": "code", + "execution_count": 21, + "metadata": {}, + "outputs": [], + "source": [ + "genus = [i.split()[0] for i in labelList]\n", + "genus_vocab=list(np.unique(genus))\n", + "y=[genus_vocab.index(x) for x in genus]\n", + "genus_vocab\n", + "\n", + "X3_df = pd.DataFrame(data=X3, columns=[\"pc1\",'pc2','pc3'], index=labelList)\n", + "\n", + "X3_df[\"genus\"]=genus\n", + "X3_df[\"y\"]=y\n", + "\n", + "genus_df=X3_df.groupby(\"genus\").agg({\"pc1\": list, \"pc2\":list,\"pc3\":list,\"y\":np.mean})" + ] + }, + { + "cell_type": "markdown", + "metadata": {}, + "source": [ + "### Eucleadian Distance Visualisation" + ] + }, + { + "cell_type": "code", + "execution_count": 22, + "metadata": {}, + "outputs": [ + { + "name": "stderr", + "output_type": "stream", + "text": [ + "/home/serge/anaconda3/envs/bio/lib/python3.6/site-packages/IPython/core/display.py:689: UserWarning:\n", + "\n", + "Consider using IPython.display.IFrame instead\n", + "\n" + ] + }, + { + "data": { + "text/html": [ + "<iframe id=\"igraph\" scrolling=\"no\" style=\"border:none;\" seamless=\"seamless\" src=\"https://plot.ly/~sergeman/8.embed\" height=\"1000px\" width=\"1000px\"></iframe>" + ], + "text/plain": [ + "<plotly.tools.PlotlyDisplay object>" + ] + }, + "execution_count": 22, + "metadata": {}, + "output_type": "execute_result" + } + ], + "source": [ + "data=[]\n", + "for g in genus_df.index:\n", + " trace = go.Scatter3d(\n", + " name = str(g),\n", + " x=genus_df.loc[g,\"pc1\"],\n", + " y=genus_df.loc[g,\"pc2\"],\n", + " z=genus_df.loc[g,\"pc3\"],\n", + " mode='markers',\n", + " marker=dict(\n", + " size=8,\n", + " color=genus_df.loc[g,\"y\"], # set color to an array/list of desired values\n", + " colorscale='Jet', # choose a colorscale\n", + " opacity=0.5)\n", + " )\n", + "\n", + " data.append(trace)\n", + " \n", + "\n", + "layout = go.Layout(\n", + " width=1000,\n", + " height=1000,\n", + " margin=dict(\n", + " l=0,\n", + " r=0,\n", + " b=0,\n", + " t=0\n", + " )\n", + ")\n", + "fig = go.Figure(data=data, layout=layout)\n", + "py.iplot(fig, filename='eucledian distance metric by genus')" + ] + }, + { + "cell_type": "markdown", + "metadata": {}, + "source": [ + "## Genome Inventory" + ] + }, + { + "cell_type": "code", + "execution_count": 23, + "metadata": {}, + "outputs": [], + "source": [ + "# DB=\"/data/genomes/GenSeq_fastas/train\"\n", + "DB='/media/serge/SharedSSD/data/genomes/ncbi-genomes-2019-04-07/bacterial genomes'" + ] + }, + { + "cell_type": "code", + "execution_count": null, + "metadata": {}, + "outputs": [], + "source": [ + "all_fastas = sequence.Dna2VecList.from_folder(DB).descriptions" + ] + }, + { + "cell_type": "code", + "execution_count": 26, + "metadata": {}, + "outputs": [], + "source": [ + "inventory = pd.DataFrame(data=[l.split()[1:3] for l in all_fastas], columns=[\"genus\",\"species\" ])" + ] + }, + { + "cell_type": "code", + "execution_count": 30, + "metadata": { + "scrolled": true + }, + "outputs": [ + { + "data": { + "text/html": [ + "<div>\n", + "<style scoped>\n", + " .dataframe tbody tr th:only-of-type {\n", + " vertical-align: middle;\n", + " }\n", + "\n", + " .dataframe tbody tr th {\n", + " vertical-align: top;\n", + " }\n", + "\n", + " .dataframe thead th {\n", + " text-align: right;\n", + " }\n", + "</style>\n", + "<table border=\"1\" class=\"dataframe\">\n", + " <thead>\n", + " <tr style=\"text-align: right;\">\n", + " <th></th>\n", + " <th>species</th>\n", + " </tr>\n", + " <tr>\n", + " <th>genus</th>\n", + " <th></th>\n", + " </tr>\n", + " </thead>\n", + " <tbody>\n", + " <tr>\n", + " <th>Escherichia</th>\n", + " <td>2239</td>\n", + " </tr>\n", + " <tr>\n", + " <th>Klebsiella</th>\n", + " <td>1718</td>\n", + " </tr>\n", + " <tr>\n", + " <th>Salmonella</th>\n", + " <td>1183</td>\n", + " </tr>\n", + " <tr>\n", + " <th>Bacillus</th>\n", + " <td>1172</td>\n", + " </tr>\n", + " <tr>\n", + " <th>Lactobacillus</th>\n", + " <td>953</td>\n", + " </tr>\n", + " <tr>\n", + " <th>Staphylococcus</th>\n", + " <td>889</td>\n", + " </tr>\n", + " <tr>\n", + " <th>Burkholderia</th>\n", + " <td>650</td>\n", + " </tr>\n", + " <tr>\n", + " <th>Enterococcus</th>\n", + " <td>626</td>\n", + " </tr>\n", + " <tr>\n", + " <th>Pseudomonas</th>\n", + " <td>613</td>\n", + " </tr>\n", + " <tr>\n", + " <th>Streptococcus</th>\n", + " <td>564</td>\n", + " </tr>\n", + " <tr>\n", + " <th>Acinetobacter</th>\n", + " <td>531</td>\n", + " </tr>\n", + " <tr>\n", + " <th>Bordetella</th>\n", + " <td>504</td>\n", + " </tr>\n", + " <tr>\n", + " <th>Vibrio</th>\n", + " <td>474</td>\n", + " </tr>\n", + " <tr>\n", + " <th>Xanthomonas</th>\n", + " <td>395</td>\n", + " </tr>\n", + " <tr>\n", + " <th>Mycobacterium</th>\n", + " <td>368</td>\n", + " </tr>\n", + " <tr>\n", + " <th>Borrelia</th>\n", + " <td>347</td>\n", + " </tr>\n", + " <tr>\n", + " <th>Campylobacter</th>\n", + " <td>339</td>\n", + " </tr>\n", + " <tr>\n", + " <th>Rhizobium</th>\n", + " <td>307</td>\n", + " </tr>\n", + " <tr>\n", + " <th>Enterobacter</th>\n", + " <td>298</td>\n", + " </tr>\n", + " <tr>\n", + " <th>Mycoplasma</th>\n", + " <td>290</td>\n", + " </tr>\n", + " <tr>\n", + " <th>Corynebacterium</th>\n", + " <td>258</td>\n", + " </tr>\n", + " <tr>\n", + " <th>Phaeobacter</th>\n", + " <td>256</td>\n", + " </tr>\n", + " <tr>\n", + " <th>Yersinia</th>\n", + " <td>243</td>\n", + " </tr>\n", + " <tr>\n", + " <th>Brucella</th>\n", + " <td>233</td>\n", + " </tr>\n", + " <tr>\n", + " <th>Clostridium</th>\n", + " <td>228</td>\n", + " </tr>\n", + " <tr>\n", + " <th>Listeria</th>\n", + " <td>227</td>\n", + " </tr>\n", + " <tr>\n", + " <th>Streptomyces</th>\n", + " <td>225</td>\n", + " </tr>\n", + " <tr>\n", + " <th>Chlamydia</th>\n", + " <td>202</td>\n", + " </tr>\n", + " <tr>\n", + " <th>Helicobacter</th>\n", + " <td>187</td>\n", + " </tr>\n", + " <tr>\n", + " <th>Candidatus</th>\n", + " <td>187</td>\n", + " </tr>\n", + " <tr>\n", + " <th>...</th>\n", + " <td>...</td>\n", + " </tr>\n", + " <tr>\n", + " <th>'Deinococcus</th>\n", + " <td>1</td>\n", + " </tr>\n", + " <tr>\n", + " <th>Mariniflexile</th>\n", + " <td>1</td>\n", + " </tr>\n", + " <tr>\n", + " <th>Marinithermus</th>\n", + " <td>1</td>\n", + " </tr>\n", + " <tr>\n", + " <th>Lactobacillales</th>\n", + " <td>1</td>\n", + " </tr>\n", + " <tr>\n", + " <th>Marinobacterium</th>\n", + " <td>1</td>\n", + " </tr>\n", + " <tr>\n", + " <th>Megamonas</th>\n", + " <td>1</td>\n", + " </tr>\n", + " <tr>\n", + " <th>Melioribacter</th>\n", + " <td>1</td>\n", + " </tr>\n", + " <tr>\n", + " <th>Melittangium</th>\n", + " <td>1</td>\n", + " </tr>\n", + " <tr>\n", + " <th>Methylobacillus</th>\n", + " <td>1</td>\n", + " </tr>\n", + " <tr>\n", + " <th>Methylocaldum</th>\n", + " <td>1</td>\n", + " </tr>\n", + " <tr>\n", + " <th>Magnetococcus</th>\n", + " <td>1</td>\n", + " </tr>\n", + " <tr>\n", + " <th>Mageeibacillus</th>\n", + " <td>1</td>\n", + " </tr>\n", + " <tr>\n", + " <th>Lysinimonas</th>\n", + " <td>1</td>\n", + " </tr>\n", + " <tr>\n", + " <th>Luteitalea</th>\n", + " <td>1</td>\n", + " </tr>\n", + " <tr>\n", + " <th>Lacunisphaera</th>\n", + " <td>1</td>\n", + " </tr>\n", + " <tr>\n", + " <th>Lautropia</th>\n", + " <td>1</td>\n", + " </tr>\n", + " <tr>\n", + " <th>Leadbetterella</th>\n", + " <td>1</td>\n", + " </tr>\n", + " <tr>\n", + " <th>Leminorella</th>\n", + " <td>1</td>\n", + " </tr>\n", + " <tr>\n", + " <th>Lentibacillus</th>\n", + " <td>1</td>\n", + " </tr>\n", + " <tr>\n", + " <th>Lentzea</th>\n", + " <td>1</td>\n", + " </tr>\n", + " <tr>\n", + " <th>Leptothrix</th>\n", + " <td>1</td>\n", + " </tr>\n", + " <tr>\n", + " <th>Levyella</th>\n", + " <td>1</td>\n", + " </tr>\n", + " <tr>\n", + " <th>Limnobaculum</th>\n", + " <td>1</td>\n", + " </tr>\n", + " <tr>\n", + " <th>Limnochorda</th>\n", + " <td>1</td>\n", + " </tr>\n", + " <tr>\n", + " <th>Litorilituus</th>\n", + " <td>1</td>\n", + " </tr>\n", + " <tr>\n", + " <th>Lonsdalea</th>\n", + " <td>1</td>\n", + " </tr>\n", + " <tr>\n", + " <th>Luteibacter</th>\n", + " <td>1</td>\n", + " </tr>\n", + " <tr>\n", + " <th>Luteipulveratus</th>\n", + " <td>1</td>\n", + " </tr>\n", + " <tr>\n", + " <th>secondary</th>\n", + " <td>1</td>\n", + " </tr>\n", + " <tr>\n", + " <th>plasmid1</th>\n", + " <td>0</td>\n", + " </tr>\n", + " </tbody>\n", + "</table>\n", + "<p>1120 rows × 1 columns</p>\n", + "</div>" + ], + "text/plain": [ + " species\n", + "genus \n", + "Escherichia 2239\n", + "Klebsiella 1718\n", + "Salmonella 1183\n", + "Bacillus 1172\n", + "Lactobacillus 953\n", + "Staphylococcus 889\n", + "Burkholderia 650\n", + "Enterococcus 626\n", + "Pseudomonas 613\n", + "Streptococcus 564\n", + "Acinetobacter 531\n", + "Bordetella 504\n", + "Vibrio 474\n", + "Xanthomonas 395\n", + "Mycobacterium 368\n", + "Borrelia 347\n", + "Campylobacter 339\n", + "Rhizobium 307\n", + "Enterobacter 298\n", + "Mycoplasma 290\n", + "Corynebacterium 258\n", + "Phaeobacter 256\n", + "Yersinia 243\n", + "Brucella 233\n", + "Clostridium 228\n", + "Listeria 227\n", + "Streptomyces 225\n", + "Chlamydia 202\n", + "Helicobacter 187\n", + "Candidatus 187\n", + "... ...\n", + "'Deinococcus 1\n", + "Mariniflexile 1\n", + "Marinithermus 1\n", + "Lactobacillales 1\n", + "Marinobacterium 1\n", + "Megamonas 1\n", + "Melioribacter 1\n", + "Melittangium 1\n", + "Methylobacillus 1\n", + "Methylocaldum 1\n", + "Magnetococcus 1\n", + "Mageeibacillus 1\n", + "Lysinimonas 1\n", + "Luteitalea 1\n", + "Lacunisphaera 1\n", + "Lautropia 1\n", + "Leadbetterella 1\n", + "Leminorella 1\n", + "Lentibacillus 1\n", + "Lentzea 1\n", + "Leptothrix 1\n", + "Levyella 1\n", + "Limnobaculum 1\n", + "Limnochorda 1\n", + "Litorilituus 1\n", + "Lonsdalea 1\n", + "Luteibacter 1\n", + "Luteipulveratus 1\n", + "secondary 1\n", + "plasmid1 0\n", + "\n", + "[1120 rows x 1 columns]" + ] + }, + "execution_count": 30, + "metadata": {}, + "output_type": "execute_result" + } + ], + "source": [ + "inventory.groupby(\"genus\").agg({\"species\":\"count\"}).sort_values(\"species\",ascending=False)" + ] + }, + { + "cell_type": "code", + "execution_count": 24, + "metadata": { + "scrolled": false + }, + "outputs": [ + { + "data": { + "text/html": [ + "<div>\n", + "<style scoped>\n", + " .dataframe tbody tr th:only-of-type {\n", + " vertical-align: middle;\n", + " }\n", + "\n", + " .dataframe tbody tr th {\n", + " vertical-align: top;\n", + " }\n", + "\n", + " .dataframe thead th {\n", + " text-align: right;\n", + " }\n", + "</style>\n", + "<table border=\"1\" class=\"dataframe\">\n", + " <thead>\n", + " <tr style=\"text-align: right;\">\n", + " <th></th>\n", + " <th></th>\n", + " <th>count</th>\n", + " </tr>\n", + " <tr>\n", + " <th>genus</th>\n", + " <th>species</th>\n", + " <th></th>\n", + " </tr>\n", + " </thead>\n", + " <tbody>\n", + " <tr>\n", + " <th>'Catharanthus</th>\n", + " <th>roseus'</th>\n", + " <td>2</td>\n", + " </tr>\n", + " <tr>\n", + " <th>'Deinococcus</th>\n", + " <th>soli'</th>\n", + " <td>1</td>\n", + " </tr>\n", + " <tr>\n", + " <th>'Nostoc</th>\n", + " <th>azollae'</th>\n", + " <td>3</td>\n", + " </tr>\n", + " <tr>\n", + " <th>18,711,729</th>\n", + " <th>reads</th>\n", + " <td>1</td>\n", + " </tr>\n", + " <tr>\n", + " <th>Acaryochloris</th>\n", + " <th>marina</th>\n", + " <td>10</td>\n", + " </tr>\n", + " <tr>\n", + " <th rowspan=\"10\" valign=\"top\">Acetobacter</th>\n", + " <th>aceti</th>\n", + " <td>1</td>\n", + " </tr>\n", + " <tr>\n", + " <th>ascendens</th>\n", + " <td>1</td>\n", + " </tr>\n", + " <tr>\n", + " <th>orientalis</th>\n", + " <td>2</td>\n", + " </tr>\n", + " <tr>\n", + " <th>oryzifermentans</th>\n", + " <td>1</td>\n", + " </tr>\n", + " <tr>\n", + " <th>pasteurianus</th>\n", + " <td>91</td>\n", + " </tr>\n", + " <tr>\n", + " <th>persici</th>\n", + " <td>2</td>\n", + " </tr>\n", + " <tr>\n", + " <th>pomorum</th>\n", + " <td>7</td>\n", + " </tr>\n", + " <tr>\n", + " <th>senegalensis</th>\n", + " <td>2</td>\n", + " </tr>\n", + " <tr>\n", + " <th>sp.</th>\n", + " <td>7</td>\n", + " </tr>\n", + " <tr>\n", + " <th>tropicalis</th>\n", + " <td>2</td>\n", + " </tr>\n", + " <tr>\n", + " <th>Acetobacterium</th>\n", + " <th>woodii</th>\n", + " <td>1</td>\n", + " </tr>\n", + " <tr>\n", + " <th>Acetohalobium</th>\n", + " <th>arabaticum</th>\n", + " <td>1</td>\n", + " </tr>\n", + " <tr>\n", + " <th>Acetomicrobium</th>\n", + " <th>mobile</th>\n", + " <td>1</td>\n", + " </tr>\n", + " <tr>\n", + " <th rowspan=\"4\" valign=\"top\">Acholeplasma</th>\n", + " <th>axanthum</th>\n", + " <td>1</td>\n", + " </tr>\n", + " <tr>\n", + " <th>hippikon</th>\n", + " <td>2</td>\n", + " </tr>\n", + " <tr>\n", + " <th>laidlawii</th>\n", + " <td>2</td>\n", + " </tr>\n", + " <tr>\n", + " <th>oculi</th>\n", + " <td>1</td>\n", + " </tr>\n", + " <tr>\n", + " <th rowspan=\"5\" valign=\"top\">Achromobacter</th>\n", + " <th>denitrificans</th>\n", + " <td>3</td>\n", + " </tr>\n", + " <tr>\n", + " <th>insolitus</th>\n", + " <td>4</td>\n", + " </tr>\n", + " <tr>\n", + " <th>sp.</th>\n", + " <td>2</td>\n", + " </tr>\n", + " <tr>\n", + " <th>spanius</th>\n", + " <td>4</td>\n", + " </tr>\n", + " <tr>\n", + " <th>xylosoxidans</th>\n", + " <td>11</td>\n", + " </tr>\n", + " <tr>\n", + " <th rowspan=\"2\" valign=\"top\">Acidaminococcus</th>\n", + " <th>fermentans</th>\n", + " <td>1</td>\n", + " </tr>\n", + " <tr>\n", + " <th>intestini</th>\n", + " <td>1</td>\n", + " </tr>\n", + " <tr>\n", + " <th>Acidiferrobacter</th>\n", + " <th>sp.</th>\n", + " <td>1</td>\n", + " </tr>\n", + " <tr>\n", + " <th>...</th>\n", + " <th>...</th>\n", + " <td>...</td>\n", + " </tr>\n", + " <tr>\n", + " <th>Zymobacter</th>\n", + " <th>palmae</th>\n", + " <td>2</td>\n", + " </tr>\n", + " <tr>\n", + " <th>Zymomonas</th>\n", + " <th>mobilis</th>\n", + " <td>49</td>\n", + " </tr>\n", + " <tr>\n", + " <th>[Arcobacter]</th>\n", + " <th>porcinus</th>\n", + " <td>1</td>\n", + " </tr>\n", + " <tr>\n", + " <th>[Bacillus</th>\n", + " <th>thuringiensis]</th>\n", + " <td>2</td>\n", + " </tr>\n", + " <tr>\n", + " <th rowspan=\"2\" valign=\"top\">[Bacillus]</th>\n", + " <th>caldolyticus</th>\n", + " <td>2</td>\n", + " </tr>\n", + " <tr>\n", + " <th>selenitireducens</th>\n", + " <td>1</td>\n", + " </tr>\n", + " <tr>\n", + " <th rowspan=\"2\" valign=\"top\">[Brevibacterium]</th>\n", + " <th>flavum</th>\n", + " <td>2</td>\n", + " </tr>\n", + " <tr>\n", + " <th>frigoritolerans</th>\n", + " <td>1</td>\n", + " </tr>\n", + " <tr>\n", + " <th rowspan=\"8\" valign=\"top\">[Clostridium]</th>\n", + " <th>bolteae</th>\n", + " <td>2</td>\n", + " </tr>\n", + " <tr>\n", + " <th>cellulolyticum</th>\n", + " <td>1</td>\n", + " </tr>\n", + " <tr>\n", + " <th>cellulosi</th>\n", + " <td>1</td>\n", + " </tr>\n", + " <tr>\n", + " <th>clariflavum</th>\n", + " <td>1</td>\n", + " </tr>\n", + " <tr>\n", + " <th>propionicum</th>\n", + " <td>1</td>\n", + " </tr>\n", + " <tr>\n", + " <th>saccharolyticum</th>\n", + " <td>1</td>\n", + " </tr>\n", + " <tr>\n", + " <th>scindens</th>\n", + " <td>1</td>\n", + " </tr>\n", + " <tr>\n", + " <th>stercorarium</th>\n", + " <td>4</td>\n", + " </tr>\n", + " <tr>\n", + " <th>[Enterobacter]</th>\n", + " <th>lignolyticus</th>\n", + " <td>1</td>\n", + " </tr>\n", + " <tr>\n", + " <th rowspan=\"3\" valign=\"top\">[Eubacterium]</th>\n", + " <th>eligens</th>\n", + " <td>3</td>\n", + " </tr>\n", + " <tr>\n", + " <th>hallii</th>\n", + " <td>1</td>\n", + " </tr>\n", + " <tr>\n", + " <th>rectale</th>\n", + " <td>1</td>\n", + " </tr>\n", + " <tr>\n", + " <th rowspan=\"2\" valign=\"top\">[Haemophilus]</th>\n", + " <th>ducreyi</th>\n", + " <td>14</td>\n", + " </tr>\n", + " <tr>\n", + " <th>parasuis</th>\n", + " <td>4</td>\n", + " </tr>\n", + " <tr>\n", + " <th>[Mycobacterium]</th>\n", + " <th>chelonae</th>\n", + " <td>2</td>\n", + " </tr>\n", + " <tr>\n", + " <th>[Pasteurella]</th>\n", + " <th>aerogenes</th>\n", + " <td>1</td>\n", + " </tr>\n", + " <tr>\n", + " <th>[Polyangium]</th>\n", + " <th>brachysporum</th>\n", + " <td>1</td>\n", + " </tr>\n", + " <tr>\n", + " <th>[Pseudomonas</th>\n", + " <th>syringae]</th>\n", + " <td>3</td>\n", + " </tr>\n", + " <tr>\n", + " <th>[Pseudomonas]</th>\n", + " <th>mesoacidophila</th>\n", + " <td>4</td>\n", + " </tr>\n", + " <tr>\n", + " <th>complete</th>\n", + " <th>chromosome</th>\n", + " <td>2</td>\n", + " </tr>\n", + " <tr>\n", + " <th>gamma</th>\n", + " <th>proteobacterium</th>\n", + " <td>1</td>\n", + " </tr>\n", + " <tr>\n", + " <th>secondary</th>\n", + " <th>endosymbiont</th>\n", + " <td>1</td>\n", + " </tr>\n", + " </tbody>\n", + "</table>\n", + "<p>3186 rows × 1 columns</p>\n", + "</div>" + ], + "text/plain": [ + " count\n", + "genus species \n", + "'Catharanthus roseus' 2\n", + "'Deinococcus soli' 1\n", + "'Nostoc azollae' 3\n", + "18,711,729 reads 1\n", + "Acaryochloris marina 10\n", + "Acetobacter aceti 1\n", + " ascendens 1\n", + " orientalis 2\n", + " oryzifermentans 1\n", + " pasteurianus 91\n", + " persici 2\n", + " pomorum 7\n", + " senegalensis 2\n", + " sp. 7\n", + " tropicalis 2\n", + "Acetobacterium woodii 1\n", + "Acetohalobium arabaticum 1\n", + "Acetomicrobium mobile 1\n", + "Acholeplasma axanthum 1\n", + " hippikon 2\n", + " laidlawii 2\n", + " oculi 1\n", + "Achromobacter denitrificans 3\n", + " insolitus 4\n", + " sp. 2\n", + " spanius 4\n", + " xylosoxidans 11\n", + "Acidaminococcus fermentans 1\n", + " intestini 1\n", + "Acidiferrobacter sp. 1\n", + "... ...\n", + "Zymobacter palmae 2\n", + "Zymomonas mobilis 49\n", + "[Arcobacter] porcinus 1\n", + "[Bacillus thuringiensis] 2\n", + "[Bacillus] caldolyticus 2\n", + " selenitireducens 1\n", + "[Brevibacterium] flavum 2\n", + " frigoritolerans 1\n", + "[Clostridium] bolteae 2\n", + " cellulolyticum 1\n", + " cellulosi 1\n", + " clariflavum 1\n", + " propionicum 1\n", + " saccharolyticum 1\n", + " scindens 1\n", + " stercorarium 4\n", + "[Enterobacter] lignolyticus 1\n", + "[Eubacterium] eligens 3\n", + " hallii 1\n", + " rectale 1\n", + "[Haemophilus] ducreyi 14\n", + " parasuis 4\n", + "[Mycobacterium] chelonae 2\n", + "[Pasteurella] aerogenes 1\n", + "[Polyangium] brachysporum 1\n", + "[Pseudomonas syringae] 3\n", + "[Pseudomonas] mesoacidophila 4\n", + "complete chromosome 2\n", + "gamma proteobacterium 1\n", + "secondary endosymbiont 1\n", + "\n", + "[3186 rows x 1 columns]" + ] + }, + "execution_count": 24, + "metadata": {}, + "output_type": "execute_result" + } + ], + "source": [ + ".groupby([\"genus\", \"species\"]).agg({\"species\": \"count\"})\n", + "inventory.columns=[\"count\"]\n", + "inventory" + ] + }, + { + "cell_type": "code", + "execution_count": 25, + "metadata": {}, + "outputs": [ + { + "data": { + "text/plain": [ + "['NZ_CP013305.1 Vibrio cholerae strain CRC1106 chromosome 1, complete sequence',\n", + " 'NZ_CP013306.1 Vibrio cholerae strain CRC1106 chromosome 2, complete sequence',\n", + " 'NC_013791.2 Bacillus pseudofirmus OF4, complete genome',\n", + " 'NC_013792.1 Bacillus pseudofirmus OF4 plasmid pBpOF4-01, complete sequence',\n", + " 'NC_013793.1 Bacillus pseudofirmus OF4 plasmid pBpOF4-02, complete sequence',\n", + " 'NC_000913.3 Escherichia coli str. K-12 substr. MG1655, complete genome',\n", + " 'NC_007164.1 Corynebacterium jeikeium K411 complete genome',\n", + " 'NC_003080.1 Corynebacterium jeikeium K411 plasmid pKW4, complete sequence',\n", + " 'NC_002162.1 Ureaplasma parvum serovar 3 str. ATCC 700970, complete genome',\n", + " 'NC_004088.1 Yersinia pestis KIM10+, complete genome',\n", + " 'NC_004838.1 Yersinia pestis KIM10+ plasmid pMT-1, complete sequence',\n", + " 'NC_002620.2 Chlamydia muridarum Nigg, complete genome',\n", + " 'NC_002182.1 Chlamydia muridarum Nigg plasmid pMoPn, complete sequence',\n", + " 'NC_002488.3 Xylella fastidiosa 9a5c, complete genome',\n", + " 'NC_002489.3 Xylella fastidiosa 9a5c plasmid pXF1.3, complete sequence',\n", + " 'NC_002490.1 Xylella fastidiosa 9a5c plasmid pXF51, complete sequence',\n", + " 'NC_002505.1 Vibrio cholerae O1 biovar El Tor str. N16961 chromosome I, complete sequence',\n", + " 'NC_002506.1 Vibrio cholerae O1 biovar El Tor str. N16961 chromosome II, complete sequence',\n", + " 'NC_002516.2 Pseudomonas aeruginosa PAO1, complete genome',\n", + " 'NC_002737.2 Streptococcus pyogenes M1 GAS, complete genome',\n", + " 'NC_002663.1 Pasteurella multocida subsp. multocida str. Pm70, complete genome',\n", + " 'NC_002946.2 Neisseria gonorrhoeae FA 1090 chromosome, complete genome',\n", + " 'NC_002662.1 Lactococcus lactis subsp. lactis Il1403 chromosome, complete genome',\n", + " 'NC_003028.3 Streptococcus pneumoniae TIGR4, complete genome',\n", + " 'NC_002696.2 Caulobacter crescentus CB15 chromosome, complete genome',\n", + " 'NC_004337.2 Shigella flexneri 2a str. 301 chromosome, complete genome',\n", + " 'NC_004851.1 Shigella flexneri 2a str. 301 plasmid pCP301, complete sequence',\n", + " 'NC_013174.1 Jonesia denitrificans DSM 20603, complete genome',\n", + " 'NC_013166.1 Kangiella koreensis DSM 16069, complete genome',\n", + " 'NC_013171.1 Anaerococcus prevotii DSM 20548, complete genome',\n", + " 'NC_013164.1 Anaerococcus prevotii DSM 20548 plasmid pAPRE01, complete sequence',\n", + " 'NC_013222.1 Robiginitalea biformata HTCC2501, complete genome',\n", + " 'NC_017285.1 Chlamydophila pneumoniae LPCoLN, complete genome',\n", + " 'NC_017286.1 Chlamydophila pneumoniae LPCoLN plasmid, complete sequence',\n", + " 'NC_013194.1 Candidatus Accumulibacter phosphatis clade IIA str. UW-1, complete genome',\n", + " 'NC_013193.1 Candidatus Accumulibacter phosphatis clade IIA str. UW-1 plasmid pAph01, complete sequence',\n", + " 'NC_013190.1 Candidatus Accumulibacter phosphatis clade IIA str. UW-1 plasmid pAph02, complete sequence',\n", + " 'NC_013191.1 Candidatus Accumulibacter phosphatis clade IIA str. UW-1 plasmid pAph03, complete sequence',\n", + " 'NC_013216.1 Desulfotomaculum acetoxidans DSM 771, complete genome',\n", + " 'NC_013203.1 Atopobium parvulum DSM 20469, complete genome',\n", + " 'NC_013355.1 Zymomonas mobilis subsp. mobilis NCIMB 11163, complete genome',\n", + " 'NC_013356.1 Zymomonas mobilis subsp. mobilis NCIMB 11163 plasmid pZA1001, complete sequence',\n", + " 'NC_013357.1 Zymomonas mobilis subsp. mobilis NCIMB 11163 plasmid pZA1002, complete sequence',\n", + " 'NC_013358.1 Zymomonas mobilis subsp. mobilis NCIMB 11163 plasmid pZA1003, complete sequence',\n", + " 'NC_013204.1 Eggerthella lenta DSM 2243, complete genome',\n", + " 'NC_013205.1 Alicyclobacillus acidocaldarius subsp. acidocaldarius DSM 446, complete genome',\n", + " 'NC_013206.1 Alicyclobacillus acidocaldarius subsp. acidocaldarius DSM 446 plasmid pAACI01, complete sequence',\n", + " 'NC_013207.1 Alicyclobacillus acidocaldarius subsp. acidocaldarius DSM 446 plasmid pAACI02, complete sequence',\n", + " 'NC_013208.1 Alicyclobacillus acidocaldarius subsp. acidocaldarius DSM 446 plasmid pAACI03, complete sequence',\n", + " 'NC_013223.1 Desulfohalobium retbaense DSM 5692, complete genome',\n", + " 'NC_013224.1 Desulfohalobium retbaense DSM 5692 plasmid pDRET01, complete sequence',\n", + " 'NC_009883.1 Rickettsia bellii OSU 85-389, complete genome',\n", + " 'NC_009953.1 Salinispora arenicola CNS-205, complete genome',\n", + " 'NC_009901.1 Shewanella pealeana ATCC 700345, complete genome',\n", + " 'NC_009922.1 Alkaliphilus oremlandii OhILAs, complete genome',\n", + " 'NC_010612.1 Mycobacterium marinum M, complete genome',\n", + " 'NC_010604.1 Mycobacterium marinum M plasmid pMM23, complete sequence',\n", + " 'NC_012125.1 Salmonella enterica subsp. enterica serovar Paratyphi C strain RKS4594, complete genome',\n", + " 'NC_012124.1 Salmonella enterica subsp. enterica serovar Paratyphi C strain RKS4594 plasmid pSPCV, complete sequence',\n", + " 'NC_009943.1 Desulfococcus oleovorans Hxd3, complete genome',\n", + " 'NC_010424.1 Candidatus Desulforudis audaxviator MP104C, complete genome',\n", + " 'NC_010611.1 Acinetobacter baumannii ACICU, complete genome',\n", + " 'NC_010605.1 Acinetobacter baumannii ACICU plasmid pACICU1, complete sequence',\n", + " 'NC_010606.1 Acinetobacter baumannii ACICU plasmid pACICU2, complete sequence',\n", + " 'NC_010084.1 Burkholderia multivorans ATCC 17616 chromosome 1, complete sequence',\n", + " 'NC_010086.1 Burkholderia multivorans ATCC 17616 chromosome 2, complete sequence',\n", + " 'NC_010087.1 Burkholderia multivorans ATCC 17616 chromosome 3, complete sequence',\n", + " 'NC_010070.1 Burkholderia multivorans ATCC 17616 plasmid pBMUL01, complete sequence',\n", + " 'NC_010103.1 Brucella canis ATCC 23365 chromosome I, complete sequence',\n", + " 'NC_010104.1 Brucella canis ATCC 23365 chromosome II, complete sequence',\n", + " 'NC_012587.1 Sinorhizobium fredii NGR234 chromosome, complete genome',\n", + " 'NC_000914.2 Sinorhizobium fredii NGR234 plasmid pNGR234a, complete sequence',\n", + " 'NC_012586.1 Sinorhizobium fredii NGR234 plasmid pNGR234b, complete sequence',\n", + " 'NC_010003.1 Petrotoga mobilis SJ95, complete genome',\n", + " 'NC_010067.1 Salmonella enterica subsp. arizonae serovar 62:z4,z23:--, complete genome',\n", + " 'NC_010002.1 Delftia acidovorans SPH-1, complete genome',\n", + " 'NC_010001.1 Lachnoclostridium phytofermentans ISDg, complete genome',\n", + " 'NC_010102.1 Salmonella enterica subsp. enterica serovar Paratyphi B str. SPB7, complete genome',\n", + " 'NC_010742.1 Brucella abortus S19 chromosome 1, complete sequence',\n", + " 'NC_010740.1 Brucella abortus S19 chromosome 2, complete sequence',\n", + " 'NC_010117.1 Coxiella burnetii RSA 331, complete genome',\n", + " 'NC_010115.1 Coxiella burnetii RSA 331 plasmid QpH1, complete sequence',\n", + " 'NC_009997.1 Shewanella baltica OS195, complete genome',\n", + " 'NC_009998.1 Shewanella baltica OS195 plasmid pS19501, complete sequence',\n", + " 'NC_009999.1 Shewanella baltica OS195 plasmid pS19502, complete sequence',\n", + " 'NC_010000.1 Shewanella baltica OS195 plasmid pS19503, complete sequence',\n", + " 'NC_010163.1 Acholeplasma laidlawii PG-8A, complete genome',\n", + " 'NC_003197.2 Salmonella enterica subsp. enterica serovar Typhimurium str. LT2, complete genome',\n", + " 'NC_003277.2 Salmonella enterica subsp. enterica serovar Typhimurium str. LT2 plasmid pSLT, complete sequence',\n", + " 'NC_004631.1 Salmonella enterica subsp. enterica serovar Typhi Ty2, complete genome',\n", + " 'NC_002944.2 Mycobacterium avium subsp. paratuberculosis str. k10, complete genome',\n", + " 'NC_007332.1 Mycoplasma hyopneumoniae 7448, complete genome',\n", + " 'NC_005957.1 [Bacillus thuringiensis] serovar konkukian str. 97-27 chromosome, complete genome',\n", + " 'NC_006578.1 [Bacillus thuringiensis] serovar konkukian str. 97-27 plasmid pBT9727, complete sequence',\n", + " 'NC_010816.1 Bifidobacterium longum DJO10A, complete genome',\n", + " 'NC_004252.1 Bifidobacterium longum DJO10A plasmid pDOJH10L, complete sequence',\n", + " 'NC_004253.1 Bifidobacterium longum DJO10A plasmid pDOJH10S, complete sequence',\n", + " 'NC_008800.1 Yersinia enterocolitica subsp. enterocolitica 8081 chromosome, complete genome',\n", + " 'NC_008791.1 Yersinia enterocolitica subsp. enterocolitica 8081 plasmid pYVe8081, complete sequence',\n", + " 'NC_003272.1 Nostoc sp. PCC 7120 DNA, complete genome',\n", + " 'NC_003276.1 Nostoc sp. PCC 7120 plasmid pCC7120alpha DNA, complete genome',\n", + " 'NC_003240.1 Nostoc sp. PCC 7120 plasmid pCC7120beta DNA, complete sequence',\n", + " 'NC_003273.1 Nostoc sp. PCC 7120 plasmid pCC7120delta DNA, complete sequence',\n", + " 'NC_003270.1 Nostoc sp. PCC 7120 plasmid pCC7120epsilon DNA, complete sequence',\n", + " 'NC_003267.1 Nostoc sp. PCC 7120 plasmid pCC7120gamma DNA, complete sequence',\n", + " 'NC_003241.1 Nostoc sp. PCC 7120 plasmid pCC7120zeta DNA, complete sequence',\n", + " 'NC_009342.1 Corynebacterium glutamicum R DNA, complete genome',\n", + " 'NC_009343.1 Corynebacterium glutamicum R plasmid pCGR1 DNA, complete sequence',\n", + " 'NC_010729.1 Porphyromonas gingivalis ATCC 33277 DNA, complete genome',\n", + " 'NC_013799.1 Hydrogenobacter thermophilus TK-6 DNA, complete genome',\n", + " 'NC_003450.3 Corynebacterium glutamicum ATCC 13032 chromosome, complete genome',\n", + " 'NC_017382.1 Helicobacter pylori 51, complete genome',\n", + " 'NC_008322.1 Shewanella sp. MR-7, complete genome',\n", + " 'NC_008320.1 Shewanella sp. MR-7, complete sequence',\n", + " 'NC_008321.1 Shewanella sp. MR-4, complete genome',\n", + " 'NC_008345.1 Shewanella frigidimarina NCIMB 400, complete genome',\n", + " 'NC_008346.1 Syntrophomonas wolfei subsp. wolfei str. Goettingen G311, complete genome',\n", + " 'NC_008347.1 Maricaulis maris MCS10, complete genome',\n", + " 'NC_008344.1 Nitrosomonas eutropha C91, complete genome',\n", + " 'NC_008341.1 Nitrosomonas eutropha C91 Plasmid1, complete sequence',\n", + " 'NC_008342.1 Nitrosomonas eutropha C91 Plasmid2, complete sequence',\n", + " 'NC_008340.1 Alkalilimnicola ehrlichii MLHE-1, complete genome',\n", + " 'NC_008570.1 Aeromonas hydrophila subsp. hydrophila ATCC 7966 chromosome, complete genome',\n", + " 'NC_008435.1 Rhodopseudomonas palustris BisA53, complete genome',\n", + " 'NC_008563.1 Escherichia coli APEC O1, complete genome',\n", + " 'NC_009837.1 Escherichia coli APEC O1 plasmid pAPEC-O1-ColBM, complete sequence',\n", + " 'NC_009838.1 Escherichia coli APEC O1 plasmid pAPEC-O1-R, complete sequence',\n", + " 'NC_008576.1 Magnetococcus marinus MC-1, complete genome',\n", + " 'NC_011566.1 Shewanella piezotolerans WP3, complete genome',\n", + " 'NC_008536.1 Candidatus Solibacter usitatus Ellin6076, complete genome',\n", + " 'NC_008711.1 Paenarthrobacter aurescens TC1, complete genome',\n", + " 'NC_008712.1 Paenarthrobacter aurescens TC1 plasmid TC1, complete sequence',\n", + " 'NC_008713.1 Paenarthrobacter aurescens TC1 plasmid TC2, complete sequence',\n", + " 'NC_008554.1 Syntrophobacter fumaroxidans MPOB, complete genome',\n", + " 'NC_008595.1 Mycobacterium avium 104, complete genome',\n", + " 'NC_008596.1 Mycobacterium smegmatis str. MC2 155 chromosome, complete genome',\n", + " 'NC_008578.1 Acidothermus cellulolyticus 11B, complete genome',\n", + " 'NC_008609.1 Pelobacter propionicus DSM 2379, complete genome',\n", + " 'NC_008607.1 Pelobacter propionicus DSM 2379 plasmid pPRO1, complete sequence',\n", + " 'NC_008608.1 Pelobacter propionicus DSM 2379 plasmid pPRO2, complete sequence',\n", + " 'NC_008600.1 Bacillus thuringiensis str. Al Hakam, complete genome',\n", + " 'NC_008598.1 Bacillus thuringiensis str. Al Hakam, plasmid pALH1, complete sequence',\n", + " 'NC_013928.1 Streptococcus mutans NN2025 DNA, complete genome',\n", + " 'NC_014215.1 Propionibacterium freudenreichii subsp. shermanii CIRM-BIA1, complete genome',\n", + " 'NC_013861.1 Legionella longbeachae NSW150, complete genome',\n", + " 'NC_014544.1 Legionella longbeachae NSW150 plasmid pLLO',\n", + " 'NC_012925.1 Streptococcus suis P1/7 complete genome',\n", + " 'NC_013949.1 Helicobacter mustelae 12198 complete genome',\n", + " 'NC_003062.2 Agrobacterium fabrum str. C58 chromosome circular, complete sequence',\n", + " 'NC_003063.2 Agrobacterium fabrum str. C58 chromosome linear, complete sequence',\n", + " 'NC_003064.2 Agrobacterium fabrum str. C58 plasmid At, complete sequence',\n", + " 'NC_003065.3 Agrobacterium fabrum str. C58 plasmid Ti, complete sequence',\n", + " 'NC_007761.1 Rhizobium etli CFN 42, complete genome',\n", + " 'NC_007762.1 Rhizobium etli CFN 42 plasmid p42a, complete sequence',\n", + " 'NC_007763.1 Rhizobium etli CFN 42 plasmid p42b, complete sequence',\n", + " 'NC_007764.1 Rhizobium etli CFN 42 plasmid p42c, complete sequence',\n", + " 'NC_007765.1 Rhizobium etli CFN 42 plasmid p42e, complete sequence',\n", + " 'NC_007766.1 Rhizobium etli CFN 42 plasmid p42f, complete sequence',\n", + " 'NC_004041.2 Rhizobium etli CFN 42 plasmid symbiotic plasmid p42d, complete sequence',\n", + " 'NC_014148.1 Planctopirus limnophila DSM 3776, complete genome',\n", + " 'NC_014149.1 Planctopirus limnophila DSM 3776 plasmid pPLIM01, complete sequence',\n", + " 'NC_014212.1 Meiothermus silvanus DSM 9946, complete genome',\n", + " 'NC_014213.1 Meiothermus silvanus DSM 9946 plasmid pMESIL01, complete sequence',\n", + " 'NC_014214.1 Meiothermus silvanus DSM 9946 plasmid pMESIL02, complete sequence',\n", + " 'NC_014171.1 Bacillus thuringiensis BMB171, complete genome',\n", + " 'NC_014172.1 Bacillus thuringiensis BMB171 plasmid pBMB171, complete sequence',\n", + " 'NC_014216.1 Desulfurivibrio alkaliphilus AHT2, complete genome',\n", + " 'NC_014158.1 Tsukamurella paurometabola DSM 20162, complete genome',\n", + " 'NC_014159.1 Tsukamurella paurometabola DSM 20162 plasmid pTpau01, complete sequence',\n", + " 'NC_014166.1 Arcobacter nitrofigilis DSM 7299, complete genome',\n", + " 'NC_014147.1 Moraxella catarrhalis BBH18, complete genome',\n", + " 'NC_014100.1 Caulobacter segnis ATCC 21756, complete genome',\n", + " 'NC_014169.1 Bifidobacterium longum subsp. longum JDM301, complete genome',\n", + " 'NC_017297.1 Clostridium botulinum F str. 230613, complete genome',\n", + " 'NC_017298.1 Clostridium botulinum F str. 230613 plasmid pCBF, complete sequence',\n", + " 'NC_014218.1 Arcanobacterium haemolyticum DSM 20595, complete genome',\n", + " 'NC_016582.1 Streptomyces bingchenggensis BCW-1, complete genome',\n", + " 'NC_014220.1 Syntrophothermus lipocalidus DSM 12680, complete genome',\n", + " 'NC_017633.1 Escherichia coli ETEC H10407, complete genome',\n", + " 'NC_017721.1 Escherichia coli ETEC H10407 p52 plasmid',\n", + " 'NC_017723.1 Escherichia coli ETEC H10407 p58 plasmid',\n", + " 'NC_017722.1 Escherichia coli ETEC H10407 p666 plasmid',\n", + " 'NC_017724.1 Escherichia coli ETEC H10407 p948 plasmid',\n", + " 'NC_017441.1 Chlamydia trachomatis Sweden2, complete genome',\n", + " 'NC_017465.1 Lactobacillus fermentum CECT 5716, complete genome',\n", + " 'NC_018588.1 Listeria monocytogenes strain SLCC2372, serotype 1/2c',\n", + " 'NC_018889.1 Listeria monocytogenes serotype 1-2c str. SLCC2372 plasmid pLM1-2cUG1 complete sequence',\n", + " 'NC_016776.1 Bacteroides fragilis 638R genome',\n", + " 'NC_016810.1 Salmonella enterica subsp. enterica serovar Typhimurium SL1344 complete genome',\n", + " 'NC_017718.1 Salmonella enterica subsp. enterica serovar Typhimurium str. SL1344 plasmid pCol1B9_SL1344 complete sequence',\n", + " 'NC_017719.1 Salmonella enterica subsp. enterica serovar Typhimurium str. SL1344 plasmid pRSF1010_SL1344 complete sequence',\n", + " 'NC_017720.1 Salmonella enterica subsp. enterica serovar Typhimurium str. SL1344 plasmid pSLT_SL1344 complete sequence',\n", + " 'NC_016809.1 Haemophilus influenzae 10810 genome',\n", + " 'NC_015964.1 Haemophilus parainfluenzae T3T1 complete genome',\n", + " 'NC_016620.1 Bacteriovorax marinus SJ genome',\n", + " 'NC_019100.1 Bacteriovorax marinus SJ plasmid pBMS1 complete sequence',\n", + " 'NC_017593.1 Streptococcus pneumoniae INV200 genome',\n", + " 'NC_017592.1 Streptococcus pneumoniae OXC141 complete genome',\n", + " 'NC_017591.1 Streptococcus pneumoniae INV104 genome',\n", + " 'NC_021006.1 Streptococcus pneumoniae SPN034156 draft genome',\n", + " 'NC_021028.1 Streptococcus pneumoniae SPN034183 draft genome',\n", + " 'NC_021026.1 Streptococcus pneumoniae SPN994038 draft genome',\n", + " 'NC_021005.1 Streptococcus pneumoniae SPN994039 draft genome',\n", + " 'NC_021003.1 Streptococcus pneumoniae SPN032672 draft genome',\n", + " 'NC_021004.1 Streptococcus pneumoniae SPN033038 draft genome',\n", + " 'NC_017519.1 Mycoplasma hyorhinis MCLD, complete genome',\n", + " 'NC_015428.1 Lactobacillus buchneri NRRL B-30929, complete genome',\n", + " 'NC_015420.1 Lactobacillus buchneri NRRL B-30929 plasmid pLBUC01, complete sequence',\n", + " 'NC_015429.1 Lactobacillus buchneri NRRL B-30929 plasmid pLBUC02, complete sequence',\n", + " 'NC_015421.1 Lactobacillus buchneri NRRL B-30929 plasmid pLBUC03, complete sequence',\n", + " 'NC_023030.2 Mycoplasma gallisepticum S6, complete genome',\n", + " 'NC_015638.1 Lacinutrix sp. 5H-3-7-4, complete genome',\n", + " 'NC_015497.1 Glaciecola sp. 4H-3-7+YE-5, complete genome',\n", + " 'NC_015498.1 Glaciecola sp. 4H-3-7+YE-5 plasmid pGLAAG01, complete sequence',\n", + " 'NC_015496.1 Dokdonia sp. 4H-3-7-5, complete genome',\n", + " 'NC_015501.1 Porphyromonas asaccharolytica DSM 20707, complete genome',\n", + " 'NC_018080.1 Pseudomonas aeruginosa DK2, complete genome',\n", + " 'NC_018081.1 Enterococcus hirae ATCC 9790, complete genome',\n", + " 'NC_015845.1 Enterococcus hirae ATCC 9790 plasmid pTG9790, complete sequence',\n", + " 'NZ_CP010978.1 Pelosinus fermentans JBW45, complete genome',\n", + " 'NC_018089.1 Streptococcus mutans GS-5, complete genome',\n", + " 'NZ_CP015923.1 Salmonella enterica subsp. enterica serovar Newport str. Levine 1, complete genome',\n", + " 'NZ_CP015924.1 Salmonella enterica subsp. enterica serovar Newport str. Levine 15, complete genome',\n", + " 'NZ_CP015876.1 Pseudomonas putida SJTE-1, complete genome',\n", + " 'NZ_CP015877.1 Pseudomonas aeruginosa SJTD-1, complete genome',\n", + " 'NZ_CP010283.1 Salmonella enterica subsp. enterica serovar Newport str. CVM 21550, complete genome',\n", + " 'NZ_CP009564.1 Salmonella enterica subsp. enterica serovar Newport str. CVM 21550 plasmid pCVM21550, complete sequence',\n", + " 'NZ_CP010281.1 Salmonella enterica subsp. enterica serovar Newport str. CVM 22513, complete genome',\n", + " 'NZ_CP009562.1 Salmonella enterica subsp. enterica serovar Newport str. CVM 22513 plasmid pCVM22513, complete sequence',\n", + " 'NZ_CP010282.1 Salmonella enterica subsp. enterica serovar Newport str. CVM 21538, complete genome',\n", + " 'NZ_CP009563.1 Salmonella enterica subsp. enterica serovar Newport str. CVM 21538 plasmid pCVM21538, complete sequence',\n", + " 'NZ_CP010279.1 Salmonella enterica subsp. enterica serovar Newport str. CVM 22425, complete genome',\n", + " 'NZ_CP009560.1 Salmonella enterica subsp. enterica serovar Newport str. CVM 22425 plasmid pCVM22425, complete sequence',\n", + " 'NZ_CP009561.1 Salmonella enterica subsp. enterica serovar Newport str. CVM N18486, complete genome',\n", + " 'NZ_CP010284.1 Salmonella enterica subsp. enterica serovar Newport str. CVM N1543, complete genome',\n", + " 'NZ_CP009570.1 Salmonella enterica subsp. enterica serovar Newport str. CVM N1543 plasmid pCVMN1543, complete sequence',\n", + " 'NZ_CP010280.1 Salmonella enterica subsp. enterica serovar Newport str. CVM 22462, complete genome',\n", + " 'NZ_CP009567.1 Salmonella enterica subsp. enterica serovar Newport str. CVM 22462 plasmid pCFSAN000934_02, complete sequence',\n", + " 'NZ_CP009568.1 Salmonella enterica subsp. enterica serovar Newport str. CVM 22462 plasmid pCFSAN000934_03, complete sequence',\n", + " 'NZ_CP009566.1 Salmonella enterica subsp. enterica serovar Newport str. CVM 22462 plasmid pCVM22462, complete sequence',\n", + " 'NC_018108.1 Taylorella equigenitalis ATCC 35865, complete genome',\n", + " 'NC_018106.1 Klebsiella michiganensis E718, complete sequence',\n", + " 'NC_021501.1 Klebsiella michiganensis strain E718 plasmid pKOX_NDM1, complete sequence',\n", + " 'NC_018107.1 Klebsiella michiganensis E718 plasmid pKOX_R1, complete sequence',\n", + " 'NC_016948.1 Mycobacterium intracellulare MOTT-64, complete genome',\n", + " 'NC_016946.1 Mycobacterium intracellulare ATCC 13950, complete genome',\n", + " 'NC_016947.1 Mycobacterium intracellulare MOTT-02, complete genome',\n", + " 'NC_017049.1 Rickettsia prowazekii str. Chernikova, complete genome',\n", + " 'NZ_CP007066.1 Fusobacterium nucleatum subsp. vincentii 3_1_27 plasmid, complete sequence',\n", + " 'NZ_CP007064.1 Fusobacterium nucleatum subsp. vincentii 3_1_27 chromosome 1, complete sequence',\n", + " 'NZ_CP007065.1 Fusobacterium nucleatum subsp. vincentii 3_1_27 chromosome 2, complete sequence',\n", + " 'NZ_CP008742.1 Pseudomonas savastanoi pv. savastanoi NCPPB 3335, complete genome',\n", + " 'NC_015678.1 Streptococcus parasanguinis ATCC 15912, complete genome',\n", + " 'NC_014643.1 Rothia dentocariosa ATCC 17931, complete genome',\n", + " 'NC_014618.1 Enterobacter lignolyticus SCF1, complete genome',\n", + " 'NC_014625.1 Ketogulonicigenium vulgare Y25, complete genome',\n", + " 'NC_014621.1 Ketogulonicigenium vulgare Y25 plasmid pYP1, complete sequence',\n", + " 'NC_014626.1 Ketogulonicigenium vulgare Y25 plasmid pYP12, complete sequence',\n", + " 'NC_017161.1 Hydrogenobacter thermophilus TK-6, complete genome',\n", + " 'NC_014616.1 Bifidobacterium bifidum S17, complete genome',\n", + " 'NC_014622.2 Paenibacillus polymyxa SC2, complete genome',\n", + " 'NC_014628.2 Paenibacillus polymyxa SC2 plasmid pSC2, complete sequence',\n", + " 'NC_017455.1 Halanaerobium praevalens DSM 2228, complete genome',\n", + " 'NC_014623.1 Stigmatella aurantiaca DW4/3-1, complete genome',\n", + " 'NC_014632.1 Ilyobacter polytropus DSM 2926, complete genome',\n", + " 'NC_014633.1 Ilyobacter polytropus DSM 2926 plasmid pILYOP01, complete sequence',\n", + " 'NC_014634.1 Ilyobacter polytropus DSM 2926 plasmid pILYOP02, complete sequence',\n", + " 'NC_017451.1 Haemophilus influenzae R2866, complete genome',\n", + " 'NC_017452.1 Haemophilus influenzae R2846, complete genome',\n", + " 'NC_015174.1 Rubinisphaera brasiliensis DSM 5305, complete genome',\n", + " 'NC_015214.1 Lactobacillus amylovorus strain 30SC, complete genome',\n", + " 'NC_015213.1 Lactobacillus amylovorus strain 30SC plasmid pRKC30SC1, complete sequence',\n", + " 'NC_015218.1 Lactobacillus amylovorus strain 30SC plasmid pRKC30SC2, complete sequence',\n", + " 'NC_017569.1 Riemerella anatipestifer RA-GD, complete genome',\n", + " 'NC_017295.1 Clostridium acetobutylicum EA 2018, complete genome',\n", + " 'NC_017296.1 Clostridium acetobutylicum EA 2018 EA2018plasmid, complete sequence',\n", + " 'NC_015470.1 Chlamydia psittaci 6BC, complete genome',\n", + " 'NC_015217.1 Chlamydia psittaci 6BC plasmid p6BC, complete sequence',\n", + " 'NZ_CP019725.1 Xanthomonas perforans 91-118 chromosome, complete genome',\n", + " 'NC_015224.1 Yersinia enterocolitica subsp. palearctica 105.5R(r), complete genome',\n", + " 'NC_015475.1 Yersinia enterocolitica subsp. palearctica 105.5R(r) plasmid 105.5R(r)p, complete sequence',\n", + " 'NC_017374.1 Helicobacter pylori 2017, complete genome',\n", + " 'NC_017381.1 Helicobacter pylori 2018, complete genome',\n", + " 'NZ_CP017125.1 Mycoplasma capricolum subsp. capripneumoniae M1601 chromosome, complete genome',\n", + " 'NC_015508.1 Agrobacterium sp. H13-3 linear chromosome, complete sequence',\n", + " 'NC_015183.1 Agrobacterium sp. H13-3 circular chromosome, complete sequence',\n", + " 'NC_015184.1 Agrobacterium sp. H13-3 plasmid pAspH13-3a, complete sequence',\n", + " 'NC_017486.1 Lactococcus lactis subsp. lactis CV56, complete genome',\n", + " 'NC_017483.1 Lactococcus lactis subsp. lactis CV56 plasmid pCV56A, complete sequence',\n", + " 'NC_017487.1 Lactococcus lactis subsp. lactis CV56 plasmid pCV56B, complete sequence',\n", + " 'NC_017484.1 Lactococcus lactis subsp. lactis CV56 plasmid pCV56C, complete sequence',\n", + " 'NC_017485.1 Lactococcus lactis subsp. lactis CV56 plasmid pCV56D, complete sequence',\n", + " 'NC_017488.1 Lactococcus lactis subsp. lactis CV56 plasmid pCV56E, complete sequence',\n", + " 'NC_017244.1 Brucella melitensis M28 chromosome 1, complete sequence',\n", + " 'NC_017245.1 Brucella melitensis M28 chromosome 2, complete sequence',\n", + " 'NC_015259.1 Polymorphum gilvum SL003B-26A1, complete genome',\n", + " 'NC_015258.1 Polymorphum gilvum SL003B-26A1 plasmid pSL003B, complete sequence',\n", + " 'NC_015277.1 Sphingobacterium sp. 21, complete genome',\n", + " 'NC_016783.1 Corynebacterium diphtheriae INCA 402, complete genome',\n", + " 'NC_016800.1 Corynebacterium diphtheriae BH8, complete genome',\n", + " 'NC_017033.1 Frateuria aurantia DSM 6220, complete genome',\n", + " 'NC_019960.1 Prevotella dentalis DSM 3688 chromosome 1, complete sequence',\n", + " 'NC_019968.1 Prevotella dentalis DSM 3688 chromosome 2, complete sequence',\n", + " 'NC_019961.1 Prevotella dentalis DSM 3688 plasmid pPREDE01, complete sequence',\n", + " 'NC_019969.1 Prevotella dentalis DSM 3688 plasmid pPREDE02, complete sequence',\n", + " 'NC_019892.1 Singulisphaera acidiphila DSM 18658, complete genome',\n", + " 'NC_019893.1 Singulisphaera acidiphila DSM 18658 plasmid pSINAC01, complete sequence',\n", + " 'NC_019894.1 Singulisphaera acidiphila DSM 18658 plasmid pSINAC02, complete sequence',\n", + " 'NC_019895.1 Singulisphaera acidiphila DSM 18658 plasmid pSINAC03, complete sequence',\n", + " 'NC_017098.1 Spirochaeta africana DSM 8902, complete genome',\n", + " 'NC_017770.1 Solitalea canadensis DSM 3403, complete genome',\n", + " 'NC_016787.1 Corynebacterium diphtheriae HC03, complete genome',\n", + " 'NC_020272.1 Bacillus amyloliquefaciens IT-45, complete genome',\n", + " 'NC_020273.1 Bacillus amyloliquefaciens IT-45 plasmid pBA45-1, whole genome shotgun sequence',\n", + " 'NC_021171.1 Bacillus sp. 1NLA3E, complete genome',\n", + " 'NZ_CP007053.1 Opitutaceae bacterium TAV5, complete genome',\n", + " 'NZ_CP007054.1 Opitutaceae bacterium TAV5 plasmid, complete sequence',\n", + " 'NC_016745.1 Oceanimonas sp. GK1, complete genome',\n", + " 'NC_016746.1 Oceanimonas sp. GK1 plasmid pOCEGK01, complete sequence',\n", + " 'NC_016747.1 Oceanimonas sp. GK1 plasmid pOCEGK02, complete sequence',\n", + " 'NZ_CP007035.1 Niabella soli DSM 19437, complete genome',\n", + " 'NC_019903.1 Desulfitobacterium dichloroeliminans LMG P-21439, complete genome',\n", + " 'NC_018017.1 Desulfitobacterium dehalogenans ATCC 51507, complete genome',\n", + " 'NC_016791.1 Clostridium sp. BNL1100, complete genome',\n", + " 'NZ_CP006932.1 Candidatus Hepatoplasma crinochetorum Av, complete genome',\n", + " 'NZ_CP006829.1 Thalassolituus oleivorans R6-15, complete genome',\n", + " 'NZ_CP007122.1 Lactobacillus paracasei N1115, complete genome',\n", + " 'NZ_CP007123.1 Lactobacillus paracasei N1115 plasmid, complete sequence',\n", + " 'NZ_CP007124.1 Lactobacillus paracasei N1115 plasmid, complete sequence',\n", + " 'NZ_CP007125.1 Lactobacillus paracasei N1115 plasmid, complete sequence',\n", + " 'NZ_CP007126.1 Lactobacillus paracasei N1115 plasmid, complete sequence',\n", + " 'NC_022568.1 Listeria monocytogenes EGD, complete genome',\n", + " 'NZ_CP007244.1 Bacillus amyloliquefaciens subsp. plantarum TrigoCor1448, complete genome',\n", + " 'NZ_CP006742.1 Bacillus anthracis str. SVA11, complete genome',\n", + " 'NZ_CP006743.1 Bacillus anthracis str. SVA11 plasmid pXO1, complete sequence',\n", + " 'NZ_CP006744.1 Bacillus anthracis str. SVA11 plasmid pXO2, complete sequence',\n", + " 'NZ_CP007183.1 Campylobacter coli RM1875, complete genome',\n", + " 'NZ_CP007185.1 Campylobacter coli RM1875 plasmid pRM1875_2.4kb, complete sequence',\n", + " 'NZ_CP007186.1 Campylobacter coli RM1875 plasmid pRM1875_3.3kb, complete sequence',\n", + " 'NZ_CP007187.1 Campylobacter coli RM1875 plasmid pRM1875_3.4kbp, complete sequence',\n", + " 'NZ_CP007184.1 Campylobacter coli RM1875 plasmid pRM1875_35kb, complete sequence',\n", + " 'NZ_CP007181.1 Campylobacter coli RM4661, complete genome',\n", + " 'NZ_CP007182.1 Campylobacter coli RM4661 plasmid pRM4661_48kbp, complete sequence',\n", + " 'NZ_CP007179.1 Campylobacter coli RM5611, complete genome',\n", + " 'NZ_CP007180.1 Campylobacter coli RM5611 plasmid pRM5611_48kb, complete sequence',\n", + " 'NZ_CP004368.1 Burkholderia pseudomallei MSHR520 chromosome 1, complete sequence',\n", + " 'NZ_CP004369.1 Burkholderia pseudomallei MSHR520 chromosome 2, complete sequence',\n", + " 'NZ_CP004373.1 Gluconobacter oxydans DSM 3504, complete genome',\n", + " 'NZ_CP006571.1 Chlamydia avium 10DC88, complete genome',\n", + " 'NZ_CP006572.1 Chlamydia avium 10DC88 plasmid p10DC88, complete sequence',\n", + " 'NZ_CP007410.1 Pseudomonas brassicacearum strain DF41, complete genome',\n", + " 'NZ_CP007436.1 Bacillus pumilus strain MTCC B6033, complete genome',\n", + " 'NZ_CP007441.1 Pseudomonas stutzeri strain 28a24, complete genome',\n", + " 'NC_022904.2 Pandoraea pnomenusa 3kgm, complete genome',\n", + " 'NC_022198.1 Corynebacterium argentoratense DSM 44202, complete genome',\n", + " 'NC_022117.1 Chlamydia trachomatis strain F/1-93, complete genome',\n", + " 'NC_022118.1 Chlamydia trachomatis strain F/6-94, complete genome',\n", + " 'NC_022873.1 Bacillus thuringiensis YBT-1518, complete genome',\n", + " 'NC_020124.1 Bacillus thuringiensis YBT-1518 plasmid pBMB0228, complete sequence',\n", + " 'NC_022874.1 Bacillus thuringiensis YBT-1518 plasmid pBMB0229, complete sequence',\n", + " 'NC_022875.1 Bacillus thuringiensis YBT-1518 plasmid pBMB0230, complete sequence',\n", + " 'NC_022876.1 Bacillus thuringiensis YBT-1518 plasmid pBMB0231, complete sequence',\n", + " 'NC_022877.1 Bacillus thuringiensis YBT-1518 plasmid pBMB0232, complete sequence',\n", + " 'NC_022882.1 Bacillus thuringiensis YBT-1518 plasmid pBMB0233, complete sequence',\n", + " 'NC_022886.1 Helicobacter pylori BM012A, complete genome',\n", + " 'NC_022911.1 Helicobacter pylori BM012S, complete genome',\n", + " 'NZ_CP017152.1 Acinetobacter baumannii DU202, complete genome',\n", + " 'NZ_CP010979.1 Pseudomonas putida S13.1.2, complete genome',\n", + " 'NC_022900.1 Zymomonas mobilis subsp. mobilis str. CP4 = NRRL B-14023, complete genome',\n", + " 'NC_022901.1 Zymomonas mobilis subsp. mobilis str. CP4 = NRRL B-14023 plasmid, complete sequence',\n", + " 'NC_022913.1 Zymomonas mobilis subsp. mobilis str. CP4 = NRRL B-14023 plasmid, complete sequence',\n", + " 'NC_022902.1 Zymomonas mobilis subsp. mobilis str. CP4 = NRRL B-14023 plasmid, complete sequence',\n", + " 'NC_022903.1 Zymomonas mobilis subsp. mobilis str. CP4 = NRRL B-14023 plasmid, complete sequence',\n", + " 'NC_022910.1 Zymomonas mobilis subsp. mobilis str. CP4 = NRRL B-14023 plasmid, complete sequence',\n", + " 'NC_022909.1 Lactobacillus johnsonii N6.2, complete genome',\n", + " 'NZ_CP009365.1 Pseudomonas soli strain SJ10, complete genome',\n", + " 'NC_022964.1 Dehalococcoides mccartyi GY50, complete genome',\n", + " 'NZ_HG738867.1 Escherichia coli str. K-12 substr. MC4100 complete genome',\n", + " 'NC_022535.1 Rhizobium sp. IRBG74 circular chromosome, complete genome',\n", + " 'NC_022545.1 Rhizobium sp. IRBG74 linear chromosome, complete genome',\n", + " 'NC_022536.1 Rhizobium sp. IRBG74 plasmid IRBL74_p, complete sequence',\n", + " 'NZ_LN794217.1 Rickettsia monacensis, strain IrR/Munich, complete genome',\n", + " 'NZ_CP011913.1 Corynebacterium ulcerans FRC58, complete genome',\n", + " 'NC_019896.1 Bacillus subtilis subsp. subtilis str. BSP1, complete genome',\n", + " 'NC_019902.2 Thioalkalivibrio nitratireducens DSM 14787, complete genome',\n", + " 'NC_019908.1 Brachyspira pilosicoli P43/6/78, complete genome',\n", + " 'NC_019904.1 Echinicola vietnamensis DSM 17526, complete genome',\n", + " 'NC_019905.1 Pseudomonas putida HB3267, complete genome',\n", + " 'NC_019906.1 Pseudomonas putida HB3267 plasmid pPC9, complete sequence',\n", + " 'NC_019907.1 Liberibacter crescens BT-1, complete genome',\n", + " 'NC_019940.1 Thioflavicoccus mobilis 8321, complete genome',\n", + " 'NC_019941.1 Thioflavicoccus mobilis 8321 plasmid pTHIMO01, complete sequence',\n", + " 'NC_019936.1 Pseudomonas stutzeri RCH2, complete genome',\n", + " 'NC_019937.1 Pseudomonas stutzeri RCH2 plasmid pPSEST01, complete sequence',\n", + " 'NC_019938.1 Pseudomonas stutzeri RCH2 plasmid pPSEST02, complete sequence',\n", + " 'NC_019939.1 Pseudomonas stutzeri RCH2 plasmid pPSEST03, complete sequence',\n", + " 'NC_019955.1 Vibrio parahaemolyticus BB22OP chromosome 1, complete sequence',\n", + " 'NC_019971.1 Vibrio parahaemolyticus BB22OP chromosome 2, complete sequence',\n", + " 'NC_019970.1 Thermoanaerobacterium thermosaccharolyticum M0795, complete genome',\n", + " 'NC_019956.1 Thermoanaerobacterium thermosaccharolyticum M0795 plasmid pTHETHE01, complete sequence',\n", + " 'NC_019966.1 Mycobacterium sp. JS623, complete genome',\n", + " 'NC_019957.1 Mycobacterium sp. JS623 plasmid pMYCSM01, complete sequence',\n", + " 'NC_019958.1 Mycobacterium sp. JS623 plasmid pMYCSM02, complete sequence',\n", + " 'NC_019959.1 Mycobacterium sp. JS623 plasmid pMYCSM03, complete sequence',\n", + " 'NC_019978.1 Halobacteroides halobius DSM 5150, complete genome',\n", + " 'NC_019673.1 Saccharothrix espanaensis DSM 44229 complete genome',\n", + " 'NC_019949.1 Mycoplasma cynos C142 complete genome',\n", + " 'NC_019954.2 Tepidanaerobacter acetatoxydans Re1 complete genome',\n", + " 'NZ_CP007528.1 Salmonella enterica subsp. enterica serovar Enteritidis str. CDC_2010K_0968, complete genome',\n", + " 'NZ_CP007529.1 Salmonella enterica subsp. enterica serovar Enteritidis str. CDC_2010K_0968 plasmid00, complete sequence',\n", + " 'NZ_CP012347.1 Salmonella enterica subsp. enterica serovar Pullorum str. ATCC 9120, complete genome',\n", + " 'NZ_CP012348.1 Salmonella enterica subsp. enterica serovar Pullorum str. ATCC 9120 plasmid pCFSAN000725_01, complete sequence',\n", + " 'NZ_CP007255.1 Rhodococcus erythropolis R138, complete genome',\n", + " 'NZ_CP007256.1 Rhodococcus erythropolis R138 plasmid pCRE138, complete sequence',\n", + " 'NZ_CP007257.1 Rhodococcus erythropolis R138 plasmid pLRE138, complete sequence',\n", + " 'NZ_CP015880.1 Ensifer adhaerens strain Casida A chromosome, complete genome',\n", + " 'NZ_CP015881.1 Ensifer adhaerens strain Casida A plasmid pCasidaAA, complete sequence',\n", + " 'NZ_CP015882.1 Ensifer adhaerens strain Casida A plasmid pCasidaAB, complete sequence',\n", + " 'NZ_CP007691.1 Brucella suis bv. 2 strain PT09143 chromosome I, complete sequence',\n", + " 'NZ_CP007692.1 Brucella suis bv. 2 strain PT09143 chromosome II, complete sequence',\n", + " 'NZ_CP007693.1 Brucella suis bv. 2 strain PT09172 chromosome I, complete sequence',\n", + " 'NZ_CP007694.1 Brucella suis bv. 2 strain PT09172 chromosome II, complete sequence',\n", + " 'NZ_CP007697.1 Brucella suis bv. 2 strain Bs364CITA chromosome I, complete sequence',\n", + " 'NZ_CP007698.1 Brucella suis bv. 2 strain Bs364CITA chromosome II, complete sequence',\n", + " 'NZ_CP007720.1 Brucella suis bv. 2 strain Bs396CITA chromosome I, complete sequence',\n", + " 'NZ_CP007721.1 Brucella suis bv. 2 strain Bs396CITA chromosome II, complete sequence',\n", + " 'NZ_CP007805.1 Haemophilus influenzae CGSHiCZ412602, complete genome',\n", + " 'NZ_CP007803.1 Mycobacterium tuberculosis K, complete genome',\n", + " 'NZ_CP006696.1 Xylella fastidiosa subsp. sandyi Ann-1, complete genome',\n", + " 'NZ_CP006697.1 Xylella fastidiosa subsp. sandyi Ann-1 plasmid unnamed1, complete sequence',\n", + " 'NZ_CP006740.1 Xylella fastidiosa MUL0034, complete genome',\n", + " 'NZ_CP006739.1 Xylella fastidiosa MUL0034 plasmid unnamed2, complete sequence',\n", + " 'NZ_CP006986.1 Rhizobium sp. IE4771 chromosome, complete genome',\n", + " 'NZ_CP006987.1 Rhizobium sp. IE4771 plasmid pRetIE4771a, complete sequence',\n", + " 'NZ_CP006988.1 Rhizobium sp. IE4771 plasmid pRetIE4771b, complete sequence',\n", + " 'NZ_CP006989.1 Rhizobium sp. IE4771 plasmid pRetIE4771c, complete sequence',\n", + " 'NZ_CP006990.1 Rhizobium sp. IE4771 plasmid pRetIE4771d, complete sequence',\n", + " 'NZ_CP006991.1 Rhizobium sp. IE4771 plasmid pRetIE4771e, complete sequence',\n", + " 'NZ_CP008696.1 Pseudomonas chlororaphis strain PA23, complete genome',\n", + " 'NZ_CP006819.1 Streptococcus thermophilus ASCC 1275, complete genome',\n", + " 'NZ_CP007695.1 Brucella suis bv. 2 strain Bs143CITA chromosome I, complete sequence',\n", + " 'NZ_CP007696.1 Brucella suis bv. 2 strain Bs143CITA chromosome II, complete sequence',\n", + " 'NC_020533.1 Staphylococcus aureus subsp. aureus ST228 complete genome, isolate 16035',\n", + " 'NC_020534.1 Staphylococcus aureus subsp. aureus ST228 plasmid pI4T8 complete sequence, isolate 16035',\n", + " 'NC_020566.1 Staphylococcus aureus subsp. aureus ST228 complete genome, isolate 16125',\n", + " 'NC_020535.1 Staphylococcus aureus subsp. aureus ST228 plasmid pI5S5 complete sequence, isolate 16125',\n", + " 'NC_020568.1 Staphylococcus aureus subsp. aureus ST228 complete genome, isolate 18583',\n", + " 'NC_020539.1 Staphylococcus aureus subsp. aureus ST228 plasmid pI8T7 complete sequence, isolate 18583',\n", + " 'NC_021173.1 Burkholderia thailandensis MSMB121 chromosome 1, complete sequence',\n", + " 'NC_021174.1 Burkholderia thailandensis MSMB121 chromosome 2, complete sequence',\n", + " 'NC_021176.1 Salmonella enterica subsp. enterica serovar Typhi str. Ty21a, complete genome',\n", + " 'NC_021175.1 Streptococcus cristatus AS 1.3089, complete genome',\n", + " 'NC_021177.1 Streptomyces fulvissimus DSM 40593, complete sequence',\n", + " 'NZ_CP010539.1 Mycoplasma pneumoniae 19294, complete genome',\n", + " 'NC_021182.1 Clostridium pasteurianum BC1, complete genome',\n", + " 'NC_021183.1 Clostridium pasteurianum BC1 plasmid pCLOPA01, complete sequence',\n", + " 'NC_021181.2 Lactobacillus acidophilus La-14, complete genome',\n", + " 'NC_021194.1 Mycobacterium tuberculosis EAI5/NITR206, complete genome',\n", + " 'NC_021191.1 Actinoplanes sp. N902-109, complete genome',\n", + " 'NC_021200.1 Mycobacterium avium subsp. paratuberculosis MAP4, complete genome',\n", + " 'NC_021213.1 Streptococcus suis TL13, complete genome',\n", + " 'NZ_CP009144.1 Sinorhizobium meliloti strain RMO17, complete genome',\n", + " 'NZ_CP009145.1 Sinorhizobium meliloti strain RMO17 plasmid pSymA, complete sequence',\n", + " 'NZ_CP009146.1 Sinorhizobium meliloti strain RMO17 plasmid pSymB, complete sequence',\n", + " 'NZ_CP009215.1 Corynebacterium ureicelerivorans strain IMMIB RIV-2301, complete genome',\n", + " 'NZ_CP009216.1 Corynebacterium ureicelerivorans strain IMMIB RIV-2301 plasmid unnamed, complete sequence',\n", + " 'NZ_CP008853.1 Bacillus anthracis strain Cvac02, complete genome',\n", + " 'NZ_CP008855.1 Bacillus sp. X1(2014), complete genome',\n", + " 'NZ_CP009278.1 Sphingobacterium sp. ML3W, complete genome',\n", + " 'NZ_CP007607.1 Bacillus thuringiensis serovar kurstaki str. YBT-1520, complete genome',\n", + " 'NZ_CP007612.1 Bacillus thuringiensis serovar kurstaki str. YBT-1520 plasmid pBMB11, complete sequence',\n", + " 'NZ_CP007608.1 Bacillus thuringiensis serovar kurstaki str. YBT-1520 plasmid pBMB2062, complete sequence',\n", + " 'NZ_CP007615.1 Bacillus thuringiensis serovar kurstaki str. YBT-1520 plasmid pBMB293, complete sequence',\n", + " 'NZ_CP007616.1 Bacillus thuringiensis serovar kurstaki str. YBT-1520 plasmid pBMB400, complete sequence',\n", + " 'NZ_CP007613.1 Bacillus thuringiensis serovar kurstaki str. YBT-1520 plasmid pBMB69, complete sequence',\n", + " 'NZ_CP007609.1 Bacillus thuringiensis serovar kurstaki str. YBT-1520 plasmid pBMB7921, complete sequence',\n", + " 'NZ_CP007610.1 Bacillus thuringiensis serovar kurstaki str. YBT-1520 plasmid pBMB8240, complete sequence',\n", + " 'NZ_CP007611.1 Bacillus thuringiensis serovar kurstaki str. YBT-1520 plasmid pBMB8513, complete sequence',\n", + " 'NZ_CP007614.1 Bacillus thuringiensis serovar kurstaki str. YBT-1520 plasmid pBMB95, complete sequence',\n", + " 'NZ_CP003424.1 Serratia sp. SCBI, complete genome',\n", + " 'NZ_CP003425.1 Serratia sp. SCBI plasmid SCBI_Pl, complete sequence',\n", + " 'NZ_CP006683.1 Melissococcus plutonius S1, complete genome',\n", + " 'NZ_CP006684.1 Melissococcus plutonius S1 plasmid pMEPL_178, complete sequence',\n", + " 'NZ_CP009239.1 Cellulophaga lytica strain HI1, complete genome',\n", + " 'NZ_CP009089.1 Salmonella enterica subsp. enterica serovar Enteritidis strain OLF-SE7-100819, complete genome',\n", + " 'NZ_CP008897.1 Enterobacter cloacae ECNIH3, complete genome',\n", + " 'NZ_CP008898.1 Enterobacter cloacae ECNIH3 plasmid pENT-576, complete sequence',\n", + " 'NZ_CP008899.1 Enterobacter cloacae ECNIH3 plasmid pENT-8a4, complete sequence',\n", + " 'NZ_CP008900.1 Enterobacter cloacae ECNIH3 plasmid pENT-d4a, complete sequence',\n", + " 'NZ_CP008901.1 Enterobacter cloacae ECNIH3 plasmid pKPC-47e, complete sequence',\n", + " 'NZ_CP009090.1 Salmonella enterica subsp. enterica serovar Enteritidis strain OLF-SE8-1021710, complete genome',\n", + " 'NZ_CP008905.1 Enterobacter cloacae ECR091, complete genome',\n", + " 'NZ_CP008906.1 Enterobacter cloacae ECR091 plasmid pENT-08e, complete sequence',\n", + " 'NZ_CP008907.1 Enterobacter cloacae ECR091 plasmid pENT-4bd, complete sequence',\n", + " 'NZ_CP008908.1 Enterobacter cloacae ECR091 plasmid pKPC-47e, complete sequence',\n", + " 'NZ_CP009091.1 Salmonella enterica subsp. enterica serovar Enteritidis strain OLF-SE9-10012, complete genome',\n", + " 'NZ_CP008943.1 Yersinia pseudotuberculosis strain ATCC 6904, complete genome',\n", + " 'NZ_CP009092.1 Salmonella enterica subsp. enterica serovar Enteritidis strain OLF-SE10-10052, complete genome',\n", + " 'NZ_CP009093.1 Salmonella enterica subsp. enterica serovar Enteritidis strain OLF-SE11-10058, complete genome',\n", + " 'NZ_CP009085.2 Salmonella enterica subsp. enterica serovar Enteritidis strain OLF-SE3-98983-4, complete genome',\n", + " 'NZ_CP009084.2 Salmonella enterica subsp. enterica serovar Enteritidis strain OLF-SE2-98984-6, complete genome',\n", + " 'NZ_CP007229.1 Mycoplasma dispar strain ATCC 27140, complete genome',\n", + " 'NZ_CP016174.1 Amycolatopsis orientalis strain B-37, complete genome',\n", + " 'NZ_CP010954.1 Sphingobium sp. YBL2, complete genome',\n", + " 'NZ_CP010955.1 Sphingobium sp. YBL2 plasmid 1pYBL2-1, complete sequence',\n", + " 'NZ_CP010956.1 Sphingobium sp. YBL2 plasmid 2pYBL2-2, complete sequence',\n", + " 'NZ_CP010957.1 Sphingobium sp. YBL2 plasmid 3pYBL2-3, complete sequence',\n", + " 'NZ_CP010958.1 Sphingobium sp. YBL2 plasmid 4pYBL2-4, complete sequence',\n", + " 'NZ_CP010959.1 Sphingobium sp. YBL2 plasmid 5pYBL2-5, complete sequence',\n", + " 'NZ_CP010960.1 Sphingobium sp. YBL2 plasmid 6pYBL2-6, complete sequence',\n", + " 'NZ_CP009677.1 Mycoplasma hominis strain AF1, complete genome',\n", + " 'NZ_CP007166.1 Xanthomonas oryzae pv. oryzae PXO86, complete genome',\n", + " 'NZ_CP010976.1 Paenibacillus sp. IHBB 10380, complete genome',\n", + " 'NZ_CP010977.1 Paenibacillus sp. IHBB 10380 plasmid, complete sequence',\n", + " 'NZ_CP011005.1 Arthrobacter sp. IHBB 11108, complete genome',\n", + " 'NZ_CP011006.1 Arthrobacter sp. IHBB 11108 plasmid pAG001, complete sequence',\n", + " 'NZ_CP011004.1 Listeria monocytogenes strain N2306, complete genome',\n", + " 'NZ_CP007699.2 Streptomyces lydicus strain A02 chromosome, complete genome',\n", + " 'NZ_LN680001.1 Bacillus sp. BS34A genome assembly, chromosome: I',\n", + " 'NZ_LM995445.1 Mycoplasma capricolum subsp. capripneumoniae genome assembly 9231-Abomsa assembly V8, chromosome : I',\n", + " 'NZ_LN554882.1 Chlamydophila abortus genome assembly CAAB7, chromosome : 1',\n", + " 'NZ_LM995446.1 Escherichia coli genome assembly EcRV308Chr, chromosome : 1',\n", + " 'NZ_LM997412.1 Peptoniphilus sp. ING2-D1G genome assembly, chromosome: I',\n", + " 'NZ_LN794158.1 Candidatus Methylopumilus turicensis genome assembly Candidatus Methylopumilus turicensis MMS-10A-171, chromosome : 1',\n", + " 'NZ_LM651928.1 Shigella flexneri genome assembly NCTC1, chromosome : 1',\n", + " 'NZ_LK936442.1 Burkholderia pseudomallei genome assembly BP_3921g, chromosome : 1',\n", + " 'NZ_LK936443.1 Burkholderia pseudomallei genome assembly BP_3921g, chromosome : 2',\n", + " 'NZ_LN614827.1 Legionella fallonii LLAP-10 genome assembly LFA, chromosome : I',\n", + " 'NZ_LN614828.1 Legionella fallonii LLAP-10 genome assembly LFA, plasmid : II',\n", + " 'NZ_LN614829.1 Legionella fallonii LLAP-10 genome assembly LFA, plasmid : III',\n", + " 'NZ_CP007686.1 Listeria monocytogenes strain L2624, complete genome',\n", + " 'NZ_CP007685.1 Listeria monocytogenes strain L2676, complete genome',\n", + " 'NZ_CP011511.1 Shigella boydii strain ATCC 9210, complete genome',\n", + " 'NZ_CP007684.1 Listeria monocytogenes strain L2626, complete genome',\n", + " 'NZ_CP011501.1 Pandoraea apista strain AU2161, complete genome',\n", + " 'NZ_CP011509.1 Archangium gephyra strain DSM 2261, complete genome',\n", + " 'NZ_CP009498.1 Endomicrobium proavitum strain Rsa215, complete genome',\n", + " 'NZ_CP011827.2 Xanthomonas citri pv. citri strain jx-6, complete genome',\n", + " 'NZ_CP013665.1 Xanthomonas citri pv. citri strain jx-6 plasmid pXAC33, complete sequence',\n", + " 'NZ_CP013664.1 Xanthomonas citri pv. citri strain jx-6 plasmid pXAC64, complete sequence',\n", + " 'NZ_CP011805.1 Altererythrobacter marensis strain KCTC 22370, complete genome',\n", + " 'NZ_CP007601.1 Staphylococcus capitis subsp. capitis strain AYP1020, complete genome',\n", + " 'NZ_CP007602.1 Staphylococcus capitis subsp. capitis strain AYP1020 plasmid pAYP1020, complete sequence',\n", + " 'NZ_CP011503.1 Burkholderia pyrrocinia strain DSM 10685 chromosome 1, complete sequence',\n", + " 'NZ_CP011504.1 Burkholderia pyrrocinia strain DSM 10685 chromosome 2, complete sequence',\n", + " 'NZ_CP011505.1 Burkholderia pyrrocinia strain DSM 10685 chromosome 3, complete sequence',\n", + " 'NZ_CP011506.1 Burkholderia pyrrocinia strain DSM 10685 plasmid p2327, complete sequence',\n", + " 'NZ_CP011770.1 Croceicoccus naphthovorans strain PQ-2, complete genome',\n", + " 'NZ_CP011771.1 Croceicoccus naphthovorans strain PQ-2 plasmid p1, complete sequence',\n", + " 'NZ_CP011772.1 Croceicoccus naphthovorans strain PQ-2 plasmid p2, complete sequence',\n", + " 'NZ_CP011807.3 Pandoraea faecigallinarum strain DSM 23572, complete genome',\n", + " 'NZ_CP011808.2 Pandoraea faecigallinarum strain DSM 23572 plasmid pPF72-1, complete sequence',\n", + " 'NZ_CP011809.2 Pandoraea faecigallinarum strain DSM 23572 plasmid pPF72-2, complete sequence',\n", + " 'NZ_CP006636.1 Escherichia coli PCN061, complete genome',\n", + " 'NZ_CP006637.1 Escherichia coli PCN061 plasmid PCN061p1, complete sequence',\n", + " 'NZ_CP006638.1 Escherichia coli PCN061 plasmid PCN061p2, complete sequence',\n", + " 'NZ_CP006639.1 Escherichia coli PCN061 plasmid PCN061p3, complete sequence',\n", + " 'NZ_CP006640.1 Escherichia coli PCN061 plasmid PCN061p4, complete sequence',\n", + " 'NZ_CP006641.1 Escherichia coli PCN061 plasmid PCN061p5, complete sequence',\n", + " 'NZ_CP006642.1 Escherichia coli PCN061 plasmid PCN061p6, complete sequence',\n", + " 'NZ_CP009743.1 Burkholderia contaminans strain MS14 chromosome 1, complete sequence',\n", + " 'NZ_CP009744.1 Burkholderia contaminans strain MS14 chromosome 2, complete sequence',\n", + " 'NZ_CP009745.1 Burkholderia contaminans strain MS14 chromosome 3, complete sequence',\n", + " 'NZ_CP011855.1 Spiroplasma atrichopogonis strain GNAT3597, complete genome',\n", + " 'NZ_CP011856.1 Spiroplasma eriocheiris strain DSM 21848, complete genome',\n", + " 'NZ_CP011798.1 Enterobacter cloacae strain UW5, complete genome',\n", + " 'NZ_CP007667.1 Neisseria meningitidis strain B6116/77, complete genome',\n", + " 'NZ_CP007668.1 Neisseria meningitidis M0579, complete genome',\n", + " 'NZ_CP011882.1 Bacillus subtilis strain TO-A JPC, complete genome',\n", + " 'NZ_CP011804.1 Clostridium carboxidivorans P7 plasmid, complete sequence',\n", + " 'NZ_CP011803.1 Clostridium carboxidivorans P7, complete genome',\n", + " 'NZ_CP005969.1 Pseudomonas syringae pv. syringae B301D, complete genome',\n", + " 'NZ_CP007584.2 Salmonella enterica subsp. enterica serovar Anatum str. USDA-ARS-USMARC-1735, complete genome',\n", + " 'NZ_CP014707.1 Salmonella enterica subsp. enterica serovar Anatum strain USMARC-1735 plasmid pSAN1-1735, complete sequence',\n", + " 'NZ_CP006003.1 Myxococcus fulvus 124B02, complete genome',\n", + " 'NZ_CP011374.1 Moraxella bovoculi strain 58069, complete genome',\n", + " 'NZ_CP011375.1 Moraxella bovoculi strain 58069 plasmid, complete sequence',\n", + " 'NZ_CP011376.1 Moraxella bovoculi strain 22581, complete genome',\n", + " 'NZ_CP011377.1 Moraxella bovoculi strain 23343, complete genome',\n", + " 'NZ_CP011378.1 Moraxella bovoculi strain 28389, complete genome',\n", + " 'NZ_CP011379.1 Moraxella bovoculi strain 33362, complete genome',\n", + " 'NZ_CP011380.2 Moraxella bovoculi strain 57922, complete genome',\n", + " 'NZ_CP011365.1 Salmonella enterica subsp. enterica serovar Typhimurium strain FORC_015, complete genome',\n", + " 'NZ_CP007497.1 Streptococcus suis strain ZY05719, complete genome',\n", + " 'NZ_CP007537.1 Streptococcus pyogenes strain AP1, complete genome',\n", + " 'NZ_CP009922.2 Streptomyces xiamenensis strain 318, complete genome',\n", + " 'NZ_CP011114.1 Paenibacillus durus ATCC 35681, complete genome',\n", + " 'NZ_CP011366.1 Salinicoccus halodurans strain H3B36, complete genome',\n", + " 'NZ_CP011398.2 Listeria monocytogenes strain CFSAN008100 chromosome, complete genome',\n", + " 'NZ_CP011399.1 Listeria monocytogenes strain CFSAN008100 plasmid pCFSAN008100, complete sequence',\n", + " 'NZ_AP012555.1 Mycobacterium avium subsp. hominissuis TH135 chromosomal DNA, complete genome',\n", + " 'NZ_AP012556.1 Mycobacterium avium subsp. hominissuis TH135 plasmid pMAH135 DNA, complete genome',\n", + " 'NZ_AP013354.1 Helicobacter pylori 26695-1 DNA, complete genome',\n", + " 'NZ_AP013356.1 Helicobacter pylori 26695-1CL DNA, complete genome',\n", + " 'NZ_AP013355.1 Helicobacter pylori 26695-1CH DNA, complete genome',\n", + " 'NZ_AP013293.1 Candidatus Sulcia muelleri PSPU DNA, complete genome',\n", + " 'NZ_AP014582.1 Bordetella bronchiseptica DNA, complete genome, strain: S798',\n", + " 'NZ_AP013294.1 Bacillus sp. OxB-1 DNA, complete genome',\n", + " 'NZ_AP014524.1 Vibrio cholerae MS6 DNA, complete genome, chromosome 1',\n", + " 'NZ_AP014525.1 Vibrio cholerae MS6 DNA, complete genome, chromosome 2',\n", + " 'NZ_AP012549.1 Cyanobacterium endosymbiont of Epithemia turgida isolate EtSB Lake Yunoko DNA, complete genome',\n", + " 'NZ_AP014646.1 Pseudomonas aeruginosa DNA, complete genome, strain: NCGM 1984',\n", + " 'NZ_AP014622.1 Pseudomonas aeruginosa DNA, complete genome, strain: NCGM 1900',\n", + " 'NZ_AP014658.1 Bifidobacterium longum DNA, complete genome, strain: 105-A',\n", + " 'NZ_AP013028.1 Wolbachia endosymbiont of Cimex lectularius DNA, complete genome',\n", + " 'NZ_CP026055.1 Aeromonas caviae strain FDAARGOS_72 chromosome, complete genome',\n", + " 'NZ_CP026052.1 Salmonella enterica strain FDAARGOS_70 chromosome, complete genome',\n", + " 'NZ_CP026053.1 Salmonella enterica strain FDAARGOS_70 plasmid unnamed1, complete sequence',\n", + " 'NZ_CP026054.1 Salmonella enterica strain FDAARGOS_70 plasmid unnamed2, complete sequence',\n", + " 'NZ_CP026051.1 Proteus mirabilis strain FDAARGOS_67 chromosome, complete genome',\n", + " 'NZ_CP026050.1 Serratia marcescens strain FDAARGOS_65 chromosome, complete genome',\n", + " 'NZ_CP026047.1 Raoultella planticola strain FDAARGOS_64 chromosome, complete genome',\n", + " 'NZ_CP026048.1 Raoultella planticola strain FDAARGOS_64 plasmid unnamed1, complete sequence',\n", + " 'NZ_CP026049.1 Raoultella planticola strain FDAARGOS_64 plasmid unnamed2, complete sequence',\n", + " 'NZ_CP026046.1 Morganella morganii strain FDAARGOS_63 chromosome, complete genome',\n", + " 'NZ_CP026045.1 Citrobacter freundii strain FDAARGOS_61 chromosome, complete genome',\n", + " 'NZ_CP026044.1 Proteus mirabilis strain FDAARGOS_60 chromosome, complete genome',\n", + " 'NZ_CP009850.1 Enterobacter cloacae strain ECNIH4, complete genome',\n", + " 'NZ_CP009851.1 Enterobacter cloacae strain ECNIH4 plasmid pENT-c88, complete sequence',\n", + " 'NZ_CP009852.1 Enterobacter cloacae strain ECNIH4 plasmid pENT-e56, complete sequence',\n", + " 'NZ_CP009853.1 Enterobacter cloacae strain ECNIH4 plasmid pKPC-860, complete sequence',\n", + " 'NZ_CP009880.1 Pantoea sp. PSNIH1, complete genome',\n", + " 'NZ_CP009881.1 Pantoea sp. PSNIH1 plasmid pKPC-1c5, complete sequence',\n", + " 'NZ_CP010325.1 Pantoea sp. PSNIH1 plasmid pPSP-057, complete sequence',\n", + " 'NZ_CP009882.1 Pantoea sp. PSNIH1 plasmid pPSP-26e, complete sequence',\n", + " 'NZ_CP010326.1 Pantoea sp. PSNIH1 plasmid pPSP-3a9, complete sequence',\n", + " 'NZ_CP009883.1 Pantoea sp. PSNIH1 plasmid pPSP-a3e, complete sequence',\n", + " 'NZ_CP009884.1 Pantoea sp. PSNIH1 plasmid pPSP-ee2, complete sequence',\n", + " 'NZ_CP009854.1 Enterobacter cloacae strain ECNIH5, complete genome',\n", + " 'NZ_CP009855.1 Enterobacter cloacae strain ECNIH5 plasmid pENT-22e, complete sequence',\n", + " 'NZ_CP009856.1 Enterobacter cloacae strain ECNIH5 plasmid pENT-784, complete sequence',\n", + " 'NZ_CP009857.1 Enterobacter cloacae strain ECNIH5 plasmid pENT-d0d, complete sequence',\n", + " 'NZ_CP009858.1 Enterobacter cloacae strain ECNIH5 plasmid pKPC-47e, complete sequence',\n", + " 'NZ_CP009859.1 Escherichia coli strain ECONIH1, complete genome',\n", + " 'NZ_CP009860.1 Escherichia coli strain ECONIH1 plasmid pECO-824, complete sequence',\n", + " 'NZ_CP009861.1 Escherichia coli strain ECONIH1 plasmid pECO-b75, complete sequence',\n", + " 'NZ_CP009862.1 Escherichia coli strain ECONIH1 plasmid pKPC-629, complete sequence',\n", + " 'NZ_CP009863.1 Klebsiella pneumoniae subsp. pneumoniae strain KPNIH29, complete genome',\n", + " 'NZ_CP009864.1 Klebsiella pneumoniae subsp. pneumoniae strain KPNIH29 plasmid pKPC-e4e, complete sequence',\n", + " 'NZ_CP009865.1 Klebsiella pneumoniae subsp. pneumoniae strain KPNIH29 plasmid pKPN-80a, complete sequence',\n", + " 'NZ_CP009866.1 Pantoea sp. PSNIH2, complete genome',\n", + " 'NZ_CP009867.1 Pantoea sp. PSNIH2 plasmid pKPC-56a, complete sequence',\n", + " 'NZ_CP009868.1 Pantoea sp. PSNIH2 plasmid pPSP-100, complete sequence',\n", + " 'NZ_CP009869.1 Pantoea sp. PSNIH2 plasmid pPSP-75c, complete sequence',\n", + " 'NZ_CP009870.1 Pantoea sp. PSNIH2 plasmid pPSP-b98, complete sequence',\n", + " 'NZ_CP009871.1 Pantoea sp. PSNIH2 plasmid pPSP-cd6, complete sequence',\n", + " 'NZ_CP012074.1 Prevotella fusca JCM 17724 strain W1435 chromosome 1, complete sequence',\n", + " 'NZ_CP012075.1 Prevotella fusca JCM 17724 strain W1435 chromosome 2, complete sequence',\n", + " 'NZ_CP012067.1 Aggregatibacter aphrophilus strain W10433, complete genome',\n", + " 'NZ_CP012072.1 Actinomyces meyeri strain W712, complete genome',\n", + " 'NZ_CP012073.1 Ottowia sp. oral taxon 894 strain W10237, complete genome',\n", + " 'NZ_CP012037.1 Francisella tularensis subsp. tularensis strain WY96, complete genome',\n", + " 'NZ_CP012328.1 Spiroplasma turonicum strain Tab4c, complete genome',\n", + " 'NZ_CP012332.1 Vulgatibacter incomptus strain DSM 27710, complete genome',\n", + " 'NZ_CP011786.1 Bifidobacterium actinocoloniiforme DSM 22766, complete genome',\n", + " 'NZ_CP011995.1 Porphyromonas gingivalis strain A7436, complete genome',\n", + " 'NZ_CP011339.1 Microcystis panniformis FACHB-1757, complete genome',\n", + " 'NZ_CP010319.1 Streptococcus agalactiae strain GBS85147, complete genome',\n", + " 'NZ_CP012299.1 Microbacterium sp. CGR1, complete genome',\n", + " 'NZ_CP011929.1 Marinobacter sp. CP1, complete genome',\n", + " 'NZ_CP012357.1 Spiroplasma litorale strain TN-1, complete genome',\n", + " 'NZ_CP012358.1 Oblitimonas alkaliphila strain B4199 chromosome, complete genome',\n", + " 'NZ_CP012359.1 Oblitimonas alkaliphila strain C6819 chromosome, complete genome',\n", + " 'NZ_CP012360.1 Oblitimonas alkaliphila strain C6918 chromosome, complete genome',\n", + " 'NZ_CP012362.1 Oblitimonas alkaliphila strain D3318 chromosome, complete genome',\n", + " 'NZ_CP012363.1 Oblitimonas alkaliphila strain E1086 chromosome, complete genome',\n", + " 'NZ_CP014773.1 Mucilaginibacter sp. PAMC 26640 chromosome, complete genome',\n", + " 'NZ_CP014772.1 Mucilaginibacter sp. PAMC 26640 plasmid unnamed, complete sequence',\n", + " 'NZ_CP014771.1 Hymenobacter sp. PAMC 26554 chromosome, complete genome',\n", + " 'NZ_CP014769.1 Hymenobacter sp. PAMC 26554 plasmid unnamed1, complete sequence',\n", + " 'NZ_CP014770.1 Hymenobacter sp. PAMC 26554 plasmid unnamed2, complete sequence',\n", + " 'NZ_CP014805.2 Elizabethkingia anophelis strain CSID_3015183678 chromosome, complete genome',\n", + " 'NZ_CP014858.1 Bacillus subtilis subsp. subtilis strain D12-5 chromosome, complete genome',\n", + " 'NZ_CP014783.1 Bacillus amyloliquefaciens strain B15 chromosome, complete genome',\n", + " 'NZ_CP014790.1 Listeria monocytogenes strain 2015TE24968 chromosome, complete genome',\n", + " 'NZ_CP015985.1 Listeria monocytogenes strain 2015TE24968 plasmid pl2015TE24968, complete sequence',\n", + " 'NZ_CP014840.1 Bacillus subtilis subsp. globigii strain ATCC 49760 chromosome, complete genome',\n", + " 'NZ_CP014784.1 Pseudomonas alcaligenes strain NEB 585, complete genome',\n", + " 'NZ_CP014835.1 Streptococcus halotolerans strain HTS9 chromosome, complete genome',\n", + " 'NZ_CP014856.1 Lysinibacillus sphaericus III(3)7, complete genome',\n", + " 'NZ_CP014857.1 Lysinibacillus sphaericus III(3)7 plasmid, complete sequence',\n", + " 'NZ_CP014847.1 Bacillus thuringiensis strain HD12, complete sequence',\n", + " 'NZ_CP014848.1 Bacillus thuringiensis strain HD12 plasmid pHD120017, complete sequence',\n", + " 'NZ_CP014849.1 Bacillus thuringiensis strain HD12 plasmid pHD120038, complete sequence',\n", + " 'NZ_CP014850.1 Bacillus thuringiensis strain HD12 plasmid pHD120039, complete sequence',\n", + " 'NZ_CP014851.1 Bacillus thuringiensis strain HD12 plasmid pHD120112, complete sequence',\n", + " 'NZ_CP014852.1 Bacillus thuringiensis strain HD12 plasmid pHD120161, complete sequence',\n", + " 'NZ_CP014853.1 Bacillus thuringiensis strain HD12 plasmid pHD120345, complete sequence',\n", + " 'NZ_CP014949.1 Enterococcus faecalis strain LD33, complete genome',\n", + " 'NZ_CP014991.1 Helicobacter himalayensis strain YS1, complete genome',\n", + " 'NZ_CP014352.1 Acidipropionibacterium acidipropionici strain ATCC 55737 chromosome, complete genome',\n", + " 'NZ_CP014353.1 Acidipropionibacterium acidipropionici strain ATCC 55737 plasmid unnamed, complete sequence',\n", + " 'NZ_CP014867.1 Pseudomonas chlororaphis isolate 189 chromosome, complete genome',\n", + " 'NZ_CP014861.1 Erysipelothrix rhusiopathiae strain GXBY-1 chromosome, complete genome',\n", + " 'NZ_CP014865.1 Rickettsia prowazekii strain Naples-1 chromosome, complete genome',\n", + " 'NZ_CP015004.1 Bacillus subtilis strain SZMC 6179J, complete genome',\n", + " 'NZ_CP015005.1 Aminobacter aminovorans strain KCTC 2477, complete genome',\n", + " 'NZ_CP015006.1 Aminobacter aminovorans strain KCTC 2477 plasmid pAA01, complete sequence',\n", + " 'NZ_CP015007.1 Aminobacter aminovorans strain KCTC 2477 plasmid pAA02, complete sequence',\n", + " 'NZ_CP015008.1 Aminobacter aminovorans strain KCTC 2477 plasmid pAA03, complete sequence',\n", + " 'NZ_CP015009.1 Aminobacter aminovorans strain KCTC 2477 plasmid pAA04, complete sequence',\n", + " 'NZ_CP010838.1 Bordetella pertussis strain H374, complete genome',\n", + " 'NZ_CP010839.1 Bordetella pertussis strain H378, complete genome',\n", + " 'NZ_CP010840.1 Bordetella pertussis strain H379, complete genome',\n", + " 'NZ_CP014029.2 Klebsiella aerogenes strain FDAARGOS_152 chromosome, complete genome',\n", + " 'NZ_CP014030.2 Citrobacter sp. FDAARGOS_156 strain FDAARGOS_155 chromosome, complete genome',\n", + " 'NZ_CP014031.2 Hafnia paralvei strain FDAARGOS_158 chromosome, complete genome',\n", + " 'NZ_CP014268.2 Escherichia coli B strain C2566, complete genome',\n", + " 'NZ_CP014269.1 Escherichia coli B strain C3029, complete genome',\n", + " 'NZ_CP014270.1 Escherichia coli K-12 strain DHB4, complete genome',\n", + " 'NZ_CP014271.1 Escherichia coli K-12 strain DHB4 plasmid F128-(DHB4), complete sequence',\n", + " 'NZ_CP014272.1 Escherichia coli K-12 strain C3026, complete genome',\n", + " 'NZ_CP014273.1 Escherichia coli K-12 strain C3026 plasmid F128-(C3026), complete sequence',\n", + " 'NZ_CP014326.1 Streptococcus mitis strain SVGS_061, complete genome',\n", + " 'NZ_CP014223.1 [Clostridium] propionicum DSM 1682, complete genome',\n", + " 'NZ_CP014279.1 Corynebacterium stationis strain ATCC 6872, complete genome',\n", + " 'NZ_CP013926.1 Alteromonas stellipolaris strain LMG 21861, complete genome',\n", + " 'NZ_CP013927.1 Alteromonas stellipolaris strain LMG 21861 plasmid pASTE61-200, complete sequence',\n", + " 'NZ_CP013932.1 Alteromonas sp. Mac1, complete genome',\n", + " 'NZ_CP014322.1 Alteromonas addita strain R10SW13, complete genome',\n", + " 'NZ_CP014274.1 Stenotrophomonas sp. KCTC 12332, complete genome',\n", + " 'NZ_CP014323.1 Alteromonas macleodii strain D7, complete genome',\n", + " 'NZ_CP014301.1 Bosea sp. PAMC 26642, complete genome',\n", + " 'NZ_CP014302.1 Bosea sp. PAMC 26642 plasmid, complete sequence',\n", + " 'NZ_CP014304.1 Hymenobacter sp. PAMC26628, complete genome',\n", + " 'NZ_CP014303.1 Hymenobacter sp. PAMC26628 plasmid, complete sequence',\n", + " 'NZ_CP013928.1 Alteromonas mediterranea strain UM8, complete genome',\n", + " 'NZ_CP013929.1 Alteromonas mediterranea strain UM8 plasmid pAMEDUM8_300, complete sequence',\n", + " 'NZ_CP014206.1 Pseudodesulfovibrio indicus strain J2 chromosome, complete genome',\n", + " 'NZ_CP005083.1 Sphingobium sp. TKS chromosome 1, complete sequence',\n", + " 'NZ_CP005084.1 Sphingobium sp. TKS chromosome 2, complete sequence',\n", + " 'NZ_CP005085.1 Sphingobium sp. TKS plasmid pTK1, complete sequence',\n", + " 'NZ_CP005086.1 Sphingobium sp. TKS plasmid pTK2, complete sequence',\n", + " 'NZ_CP005087.1 Sphingobium sp. TKS plasmid pTK3, complete sequence',\n", + " 'NZ_CP005088.1 Sphingobium sp. TKS plasmid pTK4, complete sequence',\n", + " 'NZ_CP005090.1 Sphingobium sp. TKS plasmid pTK6, complete sequence',\n", + " 'NZ_CP005091.1 Sphingobium sp. TKS plasmid pTK7, complete sequence',\n", + " 'NZ_CP005092.1 Sphingobium sp. TKS plasmid pTK8, complete sequence',\n", + " 'NZ_CP005093.1 Sphingobium sp. TKS plasmid pTK9, complete sequence',\n", + " 'NZ_CP005089.1 Sphingobium sp. TKS plasmid pTK5, complete sequence',\n", + " 'NZ_CP005188.1 Sphingobium sp. MI1205 chromosome 1, complete sequence',\n", + " 'NZ_CP005189.1 Sphingobium sp. MI1205 chromosome 2, complete sequence',\n", + " 'NZ_CP005190.1 Sphingobium sp. MI1205 plasmid pMI1, complete sequence',\n", + " 'NZ_CP005191.1 Sphingobium sp. MI1205 plasmid pMI2, complete sequence',\n", + " 'NZ_CP005192.1 Sphingobium sp. MI1205 plasmid pMI3, complete sequence',\n", + " 'NZ_CP005193.1 Sphingobium sp. MI1205 plasmid pMI4, complete sequence',\n", + " 'NZ_CP014004.1 Klebsiella pneumoniae subsp. pneumoniae strain NUHL24835, complete genome',\n", + " 'NZ_CP014005.1 Klebsiella pneumoniae subsp. pneumoniae strain NUHL24835 plasmid unnamed1, complete sequence',\n", + " 'NZ_CP014006.1 Klebsiella pneumoniae subsp. pneumoniae strain NUHL24835 plasmid unnamed2, complete sequence',\n", + " 'NZ_CP013357.1 Burkholderia oklahomensis EO147 chromosome 1, complete sequence',\n", + " 'NZ_CP013356.1 Burkholderia oklahomensis EO147 chromosome 2, complete sequence',\n", + " 'NZ_CP013358.1 Burkholderia oklahomensis C6786 chromosome 1, complete sequence',\n", + " 'NZ_CP013359.1 Burkholderia oklahomensis C6786 chromosome 2, complete sequence',\n", + " 'NZ_CP013373.1 Burkholderia sp. NRF60-BP8 chromosome 1, complete sequence',\n", + " 'NZ_CP013372.1 Burkholderia sp. NRF60-BP8 chromosome 2, complete sequence',\n", + " 'NZ_CP013374.1 Burkholderia sp. NRF60-BP8 chromosome 3, complete sequence',\n", + " 'NZ_CP013389.1 Burkholderia sp. BDU8 chromosome 1, complete sequence',\n", + " 'NZ_CP013388.1 Burkholderia sp. BDU8 chromosome 2, complete sequence',\n", + " 'NZ_CP013386.1 Burkholderia sp. BDU6 chromosome 1, complete sequence',\n", + " 'NZ_CP013387.1 Burkholderia sp. BDU6 chromosome 2, complete sequence',\n", + " 'NZ_CP013384.1 Burkholderia sp. LA-2-3-30-S1-D2 chromosome 1, complete sequence',\n", + " 'NZ_CP013383.1 Burkholderia sp. LA-2-3-30-S1-D2 chromosome 2, complete sequence',\n", + " 'NZ_CP013385.1 Burkholderia sp. LA-2-3-30-S1-D2 chromosome 3, complete sequence',\n", + " 'NZ_CP013417.1 Burkholderia sp. MSMB0266 chromosome 1, complete sequence',\n", + " 'NZ_CP013418.1 Burkholderia sp. MSMB0266 chromosome 2, complete sequence',\n", + " 'NZ_CP013419.1 Burkholderia sp. MSMB0266 plasmid pMSMB0266, complete sequence',\n", + " 'NZ_CP013424.1 Burkholderia sp. MSMB0852 chromosome 1, complete sequence',\n", + " 'NZ_CP013425.1 Burkholderia sp. MSMB0852 chromosome 2, complete sequence',\n", + " 'NZ_CP013423.1 Burkholderia sp. MSMB0852 plasmid pMSMB0852, complete sequence',\n", + " 'NZ_CP013427.1 Burkholderia sp. MSMB0856 chromosome 1, complete sequence',\n", + " 'NZ_CP013428.1 Burkholderia sp. MSMB0856 chromosome 2, complete sequence',\n", + " 'NZ_CP013429.1 Burkholderia sp. MSMB0856 chromosome 3, complete sequence',\n", + " 'NZ_CP013426.1 Burkholderia sp. MSMB0856 plasmid pMSMB0856, complete sequence',\n", + " 'NZ_CP014010.1 Klebsiella pneumoniae subsp. pneumoniae strain RJF999, complete genome',\n", + " 'NZ_CP014011.1 Klebsiella pneumoniae subsp. pneumoniae strain RJF999 plasmid pRJF999, complete sequence',\n", + " 'NZ_CP014008.1 Klebsiella pneumoniae subsp. pneumoniae strain RJF293, complete genome',\n", + " 'NZ_CP014009.1 Klebsiella pneumoniae subsp. pneumoniae strain RJF293 plasmid pRJF293, complete sequence',\n", + " 'NZ_CP013457.1 Burkholderia sp. MSMB617WGS chromosome 1, complete sequence',\n", + " 'NZ_CP013458.1 Burkholderia sp. MSMB617WGS chromosome 2, complete sequence',\n", + " 'NZ_CP013992.1 Flavobacterium columnare strain 94-081, complete genome',\n", + " 'NZ_CP013920.1 Arsenophonus symbiont of Lipoptena fortisetosa strain CB, complete genome',\n", + " 'NZ_CP013975.1 Piscirickettsia salmonis strain CGR02, complete genome',\n", + " 'NZ_CP013976.1 Piscirickettsia salmonis strain CGR02 plasmid pPSCRG02-1, complete sequence',\n", + " 'NZ_CP013977.1 Piscirickettsia salmonis strain CGR02 plasmid pPSCRG02-2, complete sequence',\n", + " 'NZ_CP013978.1 Piscirickettsia salmonis strain CGR02 plasmid pPSCRG02-4, complete sequence',\n", + " 'NZ_CP013997.1 Pseudomonas monteilii strain USDA-ARS-USMARC-56711, complete genome',\n", + " 'NC_022998.1 Spiroplasma apis B31, complete genome',\n", + " 'NC_023037.2 Paenibacillus polymyxa CR1, complete genome',\n", + " 'NZ_CP006579.1 Aeromonas hydrophila 4AK4, complete genome',\n", + " 'NZ_CP006859.1 Listeria monocytogenes serotype 1/2a str. 08-6997, complete genome',\n", + " 'NZ_CP003915.1 Advenella mimigardefordensis DPN7, complete genome',\n", + " 'NZ_CP003916.1 Advenella mimigardefordensis DPN7 plasmid 24p, complete sequence',\n", + " 'NZ_CP007128.1 Gemmatirosa kalamazoonesis strain KBS708, complete genome',\n", + " 'NZ_CP007129.1 Gemmatirosa kalamazoonesis strain KBS708 plasmid 1, complete sequence',\n", + " 'NZ_CP007127.1 Gemmatirosa kalamazoonesis strain KBS708 phage-like extrachromosal element, complete sequence',\n", + " 'NZ_CP007130.1 Gemmatirosa kalamazoonesis strain KBS708 plasmid 2, complete sequence',\n", + " 'NZ_CP006720.1 Spiroplasma mirum ATCC 29335 strain SMCA, complete genome',\n", + " 'NZ_CP006715.1 Bifidobacterium breve 689b, complete genome',\n", + " 'NZ_CP007230.1 Yersinia similis strain 228, complete genome',\n", + " 'NZ_CP007231.1 Yersinia similis strain 228 plasmid, complete sequence',\n", + " 'NC_022107.1 Chlamydia trachomatis F/11-96, complete genome',\n", + " 'NZ_CP007446.1 Snodgrassella alvi wkB2, complete genome',\n", + " 'NZ_CP012371.1 Nitrosospira briensis C-128, complete genome',\n", + " 'NZ_CP010493.1 Campylobacter jejuni strain CJ677CC062, complete genome',\n", + " 'NZ_CP010494.1 Campylobacter jejuni strain CJ677CC059, complete genome',\n", + " 'NZ_CP010496.1 Campylobacter jejuni strain CJ677CC032, complete genome',\n", + " 'NZ_CP010497.1 Campylobacter jejuni strain CJ677CC033, complete genome',\n", + " 'NZ_CP010498.1 Campylobacter jejuni strain CJ677CC537, complete genome',\n", + " 'NZ_CP010499.1 Campylobacter jejuni strain CJ677CC542, complete genome',\n", + " 'NZ_CP010500.1 Campylobacter jejuni strain CJ677CC528, complete genome',\n", + " 'NZ_CP010495.1 Campylobacter jejuni strain CJ677CC538, complete genome',\n", + " 'NZ_CP010501.1 Campylobacter jejuni strain CJ677CC520, complete genome',\n", + " 'NZ_CP010502.1 Campylobacter jejuni strain CJ677CC014, complete genome',\n", + " 'NZ_CP010503.1 Campylobacter jejuni strain CJ677CC039, complete genome',\n", + " 'NZ_CP010504.1 Campylobacter jejuni strain CJ677CC085, complete genome',\n", + " 'NZ_CP010505.1 Campylobacter jejuni strain CJ677CC052, complete genome',\n", + " 'NZ_CP010506.1 Campylobacter jejuni strain CJ677CC527, complete genome',\n", + " 'NZ_CP010507.1 Campylobacter jejuni strain CJ677CC078, complete genome',\n", + " 'NZ_CP010508.1 Campylobacter jejuni strain CJ677CC523, complete genome',\n", + " 'NZ_CP010509.1 Campylobacter jejuni strain CJ677CC540, complete genome',\n", + " 'NZ_CP010510.1 Campylobacter jejuni strain CJ677CC040, complete genome',\n", + " 'NZ_CP010511.1 Campylobacter jejuni strain CJ677CC061, complete genome',\n", + " 'NZ_CP010457.1 Campylobacter jejuni strain CJ677CC539, complete genome',\n", + " 'NZ_CP010458.1 Campylobacter jejuni strain CJ677CC533, complete genome',\n", + " 'NZ_CP010459.1 Campylobacter jejuni strain CJ677CC047, complete genome',\n", + " 'NZ_CP009802.1 Streptomyces sp. FR-008, complete genome',\n", + " 'NZ_CP009803.1 Streptomyces sp. FR-008 plasmid pSSFR1, complete sequence',\n", + " 'NZ_CP009804.1 Streptomyces sp. FR-008 plasmid pSSFR2, complete sequence',\n", + " 'NZ_CP012482.1 Bacillus cellulasensis strain NJ-V2, complete genome',\n", + " 'NZ_CP013114.1 Staphylococcus equorum strain KS1039, complete genome',\n", + " 'NZ_CP009847.1 Vibrio parahaemolyticus strain FORC_004 chromosome 1, complete sequence',\n", + " 'NZ_CP009848.1 Vibrio parahaemolyticus strain FORC_004 chromosome 2, complete sequence',\n", + " 'NZ_CP009849.1 Vibrio parahaemolyticus strain FORC_004 plasmid pFORC4, complete sequence',\n", + " 'NZ_CP009984.1 Vibrio vulnificus strain FORC_009 chromosome 1, complete sequence',\n", + " 'NZ_CP009985.1 Vibrio vulnificus strain FORC_009 chromosome 2, complete sequence',\n", + " 'NZ_CP012837.1 Corynebacterium pseudotuberculosis strain 1002B, complete genome',\n", + " 'NZ_CP012907.1 Helicobacter pylori strain 29CaP, complete genome',\n", + " 'NZ_CP012905.1 Helicobacter pylori strain 7C, complete sequence',\n", + " 'NZ_CP012906.1 Helicobacter pylori strain 7C plasmid, complete sequence',\n", + " 'NZ_CP013120.1 Alteromonas stellipolaris LMG 21856, complete genome',\n", + " 'NZ_CP013121.1 Fusobacterium nucleatum subsp. polymorphum strain ChDC F306, complete genome',\n", + " 'NZ_CP013122.1 Fusobacterium nucleatum subsp. polymorphum strain ChDC F306 plasmid unnamed1, complete sequence',\n", + " 'NZ_CP013123.1 Fusobacterium nucleatum subsp. polymorphum strain ChDC F306 plasmid unnamed2, complete sequence',\n", + " 'NZ_CP014140.1 Aneurinibacillus sp. XH2, complete genome',\n", + " 'NZ_CP013126.1 Acidipropionibacterium acidipropionici strain CGMCC 1.2230 chromosome, complete genome',\n", + " 'NZ_CP012921.1 Salmonella enterica subsp. enterica serovar Heidelberg strain SA02DT10168701, complete genome',\n", + " 'NZ_CP012922.1 Salmonella enterica subsp. enterica serovar Heidelberg strain SA02DT10168701 plasmid pSA02DT10168701_37, complete sequence',\n", + " 'NZ_CP012923.1 Salmonella enterica subsp. enterica serovar Heidelberg strain SA02DT10168701 plasmid pSA02DT10168701_99, complete sequence',\n", + " 'NZ_CP012924.1 Salmonella enterica subsp. enterica serovar Heidelberg strain 12-4374, complete genome',\n", + " 'NZ_CP012925.1 Salmonella enterica subsp. enterica serovar Heidelberg strain 12-4374 plasmid p12-4374_2, complete sequence',\n", + " 'NZ_CP012927.1 Salmonella enterica subsp. enterica serovar Heidelberg strain 12-4374 plasmid p12-4374_3, complete sequence',\n", + " 'NZ_CP012926.1 Salmonella enterica subsp. enterica serovar Heidelberg strain 12-4374 plasmid p12-4374_37, complete sequence',\n", + " 'NZ_CP012928.1 Salmonella enterica subsp. enterica serovar Heidelberg strain 12-4374 plasmid p12-4374_62, complete sequence',\n", + " 'NZ_CP012929.1 Salmonella enterica subsp. enterica serovar Heidelberg strain 12-4374 plasmid p12-4374_96, complete sequence',\n", + " 'NZ_CP012930.1 Salmonella enterica subsp. enterica serovar Heidelberg strain N13-01290, complete genome',\n", + " 'NZ_CP012932.1 Salmonella enterica subsp. enterica serovar Heidelberg strain N13-01290 plasmid pN13-01290_2, complete sequence',\n", + " 'NZ_CP012931.1 Salmonella enterica subsp. enterica serovar Heidelberg strain N13-01290 plasmid pN13-01290_23, complete sequence',\n", + " 'NZ_CP012933.1 Salmonella enterica subsp. enterica serovar Heidelberg strain N13-01290 plasmid pN13-01290_3-1, complete sequence',\n", + " 'NZ_CP012934.1 Salmonella enterica subsp. enterica serovar Heidelberg strain N13-01290 plasmid pN13-01290_3-2, complete sequence',\n", + " 'NZ_CP012935.1 Salmonella enterica subsp. enterica serovar Heidelberg strain N13-01290 plasmid pN13-01290_3-3, complete sequence',\n", + " 'NZ_CP012936.1 Salmonella enterica subsp. enterica serovar Heidelberg strain N13-01290 plasmid pN13-01290_98, complete sequence',\n", + " 'NZ_CP013133.1 Rickettsia rhipicephali strain HJ#5, complete genome',\n", + " 'NZ_CP013134.1 Rickettsia rhipicephali strain HJ#5 plasmid pHJ51, complete sequence',\n", + " 'NZ_CP013135.1 Rickettsia rhipicephali strain HJ#5 plasmid pHJ52 sequence',\n", + " 'NZ_CP013112.1 Escherichia coli strain YD786, complete genome',\n", + " 'NZ_CP013141.1 Lysobacter antibioticus strain ATCC 29479 genome',\n", + " 'NZ_CP011129.1 Lysobacter antibioticus strain 76, complete genome',\n", + " 'NZ_CP012096.1 Serratia plymuthica strain 3Rp8, complete genome',\n", + " 'NZ_CP012097.1 Serratia plymuthica strain 3Re4-18, complete genome',\n", + " 'NZ_CP015963.1 Altererythrobacter ishigakiensis strain NBRC 107699 chromosome, complete genome',\n", + " 'NZ_CP016033.1 Porphyrobacter neustonensis strain DSM 9434, complete genome',\n", + " 'NZ_CP015990.1 Klebsiella pneumoniae strain BR chromosome, complete genome',\n", + " 'NZ_CP015991.1 Klebsiella pneumoniae strain BR plasmid pWSZBR, complete sequence',\n", + " 'NZ_CP014762.1 Klebsiella pneumoniae strain KPNIH39, complete genome',\n", + " 'NZ_CP014763.1 Klebsiella pneumoniae strain KPNIH39 plasmid pKPN-332, complete sequence',\n", + " 'NZ_CP014764.1 Klebsiella pneumoniae strain KPNIH39 plasmid pKPN-704, complete sequence',\n", + " 'NZ_CP014765.1 Klebsiella pneumoniae strain KPNIH39 plasmid pKpQIL-9b8, complete sequence',\n", + " 'NZ_CP012968.1 Staphylococcus sp. AntiMn-1, complete genome',\n", + " 'NZ_CP013300.1 Staphylococcus sp. AntiMn-1 plasmid AM1_232, complete sequence',\n", + " 'NZ_CP015850.1 Ralstonia solanacearum strain YC40-M chromosome, complete genome',\n", + " 'NZ_CP015851.1 Ralstonia solanacearum strain YC40-M plasmid, complete sequence',\n", + " 'NZ_CP015025.1 Klebsiella pneumoniae strain Kpn223, complete genome',\n", + " 'NZ_CP015026.1 Klebsiella pneumoniae strain Kpn223 plasmid pKPN-065, complete sequence',\n", + " 'NZ_CP015130.1 Klebsiella pneumoniae strain Kpn555 chromosome, complete genome',\n", + " 'NZ_CP015131.1 Klebsiella pneumoniae strain Kpn555 plasmid pKPN-7c3, complete sequence',\n", + " 'NZ_CP015133.1 Klebsiella pneumoniae strain Kpn555 plasmid pKPN-d6b, complete sequence',\n", + " 'NZ_CP015132.1 Klebsiella pneumoniae strain Kpn555 plasmid pKPN-d90, complete sequence',\n", + " 'NZ_CP015159.1 Escherichia coli strain Eco889 chromosome, complete genome',\n", + " 'NZ_CP015161.1 Escherichia coli strain Eco889 plasmid pECO-93a, complete sequence',\n", + " 'NZ_CP015160.1 Escherichia coli strain Eco889 plasmid pECO-fce, complete sequence',\n", + " 'NZ_CP015309.1 Corynebacterium pseudotuberculosis strain PA02 chromosome, complete genome',\n", + " 'NZ_CP014623.1 Lactobacillus backii strain TMW 1.1988 chromosome, complete genome',\n", + " 'NZ_CP014624.1 Lactobacillus backii strain TMW 1.1988 plasmid L11988-1, complete sequence',\n", + " 'NZ_CP014625.1 Lactobacillus backii strain TMW 1.1988 plasmid L11988-10, complete sequence',\n", + " 'NZ_CP014626.1 Lactobacillus backii strain TMW 1.1988 plasmid L11988-2, complete sequence',\n", + " 'NZ_CP014627.1 Lactobacillus backii strain TMW 1.1988 plasmid L11988-3, complete sequence',\n", + " 'NZ_CP014628.1 Lactobacillus backii strain TMW 1.1988 plasmid L11988-4, complete sequence',\n", + " 'NZ_CP014629.1 Lactobacillus backii strain TMW 1.1988 plasmid L11988-5, complete sequence',\n", + " 'NZ_CP014630.1 Lactobacillus backii strain TMW 1.1988 plasmid L11988-6, complete sequence',\n", + " 'NZ_CP014631.1 Lactobacillus backii strain TMW 1.1988 plasmid L11988-7, complete sequence',\n", + " 'NZ_CP014632.1 Lactobacillus backii strain TMW 1.1988 plasmid L11988-8, complete sequence',\n", + " 'NZ_CP014633.1 Lactobacillus backii strain TMW 1.1988 plasmid L11988-9, complete sequence',\n", + " 'NZ_CP012969.1 Psychrobacter sp. AntiMn-1, complete genome',\n", + " 'NZ_CP013101.1 Clostridium perfringens strain FORC_025, complete genome',\n", + " 'NZ_CP017715.1 Marinobacter salinus strain Hb8 chromosome, complete genome',\n", + " 'NZ_CP017188.2 Xanthomonas citri pv. glycines str. 8ra chromosome, complete genome',\n", + " 'NZ_CP017189.2 Xanthomonas citri pv. glycines str. 8ra plasmid pXAG27.8ra, complete sequence',\n", + " 'NZ_CP017190.1 Xanthomonas campestris pv. vesicatoria str. 85-10, complete sequence',\n", + " 'NZ_CP017191.1 Xanthomonas campestris pv. vesicatoria str. 85-10 plasmid p_XCV_1, complete sequence',\n", + " 'NZ_CP017192.1 Xanthomonas campestris pv. vesicatoria str. 85-10 plasmid p_XCV_2, complete sequence',\n", + " 'NZ_CP017193.1 Xanthomonas campestris pv. vesicatoria str. 85-10 plasmid p_XCV_3, complete sequence',\n", + " 'NZ_CP017603.1 Clostridium formicaceticum strain ATCC 27076, complete genome',\n", + " 'NZ_CP015208.1 Candidatus Rhodoluna planktonica strain MWH-Dar1, complete genome',\n", + " 'NZ_CP015381.1 Desulfococcus multivorans strain DSM 2059, complete genome',\n", + " 'NZ_CP017751.1 Cupriavidus sp. USMAHM13 chromosome 1, complete sequence',\n", + " 'NZ_CP017752.1 Cupriavidus sp. USMAHM13 chromosome 2, complete sequence',\n", + " 'NZ_CP017753.1 Cupriavidus sp. USMAHM13 plasmid unnamed1, complete sequence',\n", + " 'NZ_CP017748.1 Cupriavidus sp. USMAA2-4 chromosome 1, complete sequence',\n", + " 'NZ_CP017749.1 Cupriavidus sp. USMAA2-4 chromosome 2, complete sequence',\n", + " 'NZ_CP017750.1 Cupriavidus sp. USMAA2-4 plasmid unnamed1, complete sequence',\n", + " 'NZ_CP017754.1 Cupriavidus sp. USMAA1020 chromosome 1, complete sequence',\n", + " 'NZ_CP017755.1 Cupriavidus sp. USMAA1020 chromosome 2, complete sequence',\n", + " 'NZ_CP017756.1 Cupriavidus sp. USMAA1020 plasmid unnamed1, complete sequence',\n", + " 'NZ_CP017747.1 Bacillus velezensis strain SYBC H47 chromosome, complete genome',\n", + " 'NZ_CP017421.1 Arthrobacter sp. ZXY-2 chromosome, complete genome',\n", + " 'NZ_CP017422.1 Arthrobacter sp. ZXY-2 plasmid pZXY21, complete sequence',\n", + " 'NZ_CP017423.1 Arthrobacter sp. ZXY-2 plasmid pZXY22, complete sequence',\n", + " 'NZ_CP017424.1 Arthrobacter sp. ZXY-2 plasmid pZXY23, complete sequence',\n", + " 'NZ_CP017425.1 Arthrobacter sp. ZXY-2 plasmid pZXY24, complete sequence',\n", + " 'NZ_CP017426.1 Arthrobacter sp. ZXY-2 plasmid pZXY25, complete sequence',\n", + " 'NZ_CP017640.1 Methylobacterium sp. C1 chromosome, complete genome',\n", + " 'NZ_CP017718.1 Hyphomonas sp. Mor2 genome',\n", + " 'NZ_CP013013.1 Vibrio cholerae strain Env-390 chromosome 1, complete sequence',\n", + " 'NZ_CP013014.1 Vibrio cholerae strain Env-390 chromosome 2, complete sequence',\n", + " 'NZ_CP013475.1 Mycobacterium tuberculosis strain 1458, complete genome',\n", + " 'NZ_CP017707.1 Chromobacterium vaccinii strain 21-1 chromosome, complete genome',\n", + " 'NZ_CP017622.1 Ketogulonicigenium vulgare strain SPU B805 chromosome, complete genome',\n", + " 'NZ_CP012426.1 Klebsiella pneumoniae strain KP5 chromosome, complete genome',\n", + " 'NZ_CP012427.1 Klebsiella pneumoniae strain KP5 plasmid pSg1-1, complete sequence',\n", + " 'NZ_CP012428.1 Klebsiella pneumoniae strain KP5 plasmid pSg1-2, complete sequence',\n", + " 'NZ_CP012429.1 Klebsiella pneumoniae strain KP5 plasmid pSg1-3, complete sequence',\n", + " 'NZ_CP011839.1 Klebsiella pneumoniae strain KP5 plasmid pSg1-NDM, complete sequence',\n", + " 'NZ_CP017100.1 Escherichia coli strain K-12 NEB 5-alpha chromosome, complete genome',\n", + " 'NZ_CP017107.1 Lactobacillus salivarius strain CICC 23174 chromosome, complete genome',\n", + " 'NZ_CP017108.1 Lactobacillus salivarius strain CICC23174 plasmid pLS_1 sequence',\n", + " 'NZ_CP017109.1 Lactobacillus salivarius strain CICC23174 plasmid pLS_2 sequence',\n", + " 'NZ_CP017110.1 Lactobacillus salivarius strain CICC23174 plasmid pLS_3, complete sequence',\n", + " 'NZ_CP017124.1 Lactobacillus curvatus strain WiKim38, complete genome',\n", + " 'NZ_CP016395.1 Bacillus velezensis strain M75 chromosome, complete genome',\n", + " 'NZ_CP017111.1 Sulfurospirillum halorespirans DSM 13726 chromosome, complete genome',\n", + " 'NZ_CP017116.1 Erysipelothrix rhusiopathiae strain WH13013 chromosome, complete genome',\n", + " 'NZ_CP012692.1 Staphylococcus aureus strain FORC_027, complete genome',\n", + " 'NZ_CP016019.1 Bifidobacterium longum subsp. longum strain AH1206, complete genome',\n", + " 'NZ_CP017247.1 Bacillus licheniformis strain BL1202, complete genome',\n", + " 'NZ_CP016385.1 Salmonella enterica subsp. enterica serovar Typhimurium strain 13-931, complete genome',\n", + " 'NZ_CP016386.1 Salmonella enterica subsp. enterica serovar Typhimurium strain 13-931 plasmid p931-3904, complete sequence',\n", + " 'NZ_CP016387.1 Salmonella enterica subsp. enterica serovar Typhimurium strain 13-931 plasmid p931IncI1, complete sequence',\n", + " 'NZ_CP016388.1 Salmonella enterica subsp. enterica serovar Typhimurium strain 13-931 plasmid p931IncI2, complete sequence',\n", + " 'NZ_CP016389.1 Salmonella enterica subsp. enterica serovar Typhimurium strain 13-931 plasmid pESBL931, complete sequence',\n", + " 'NZ_CP016390.1 Salmonella enterica subsp. enterica serovar Typhimurium strain 13-931 plasmid pSLT931, complete sequence',\n", + " 'NZ_CP015217.1 Leptospira alstonii strain GWTS #1 chromosome 1, complete sequence',\n", + " 'NZ_CP015218.1 Leptospira alstonii strain GWTS #1 chromosome 2, complete sequence',\n", + " 'NZ_CP016044.1 Edwardsiella piscicida strain S11-285 chromosome, complete genome',\n", + " 'NZ_CP016445.1 Edwardsiella piscicida strain S11-285 plasmid unnamed1, complete sequence',\n", + " 'NZ_CP017157.1 Streptomyces lydicus strain 103 chromosome, complete genome',\n", + " 'NZ_CP017149.1 Pseudomonas aeruginosa strain ATCC 15692 chromosome, complete genome',\n", + " 'NZ_CP017150.1 Brevibacterium aurantiacum strain SMQ-1335 chromosome, complete genome',\n", + " 'NZ_CP017175.1 Burkholderia mallei strain Bahrain1 chromosome 1, complete sequence',\n", + " 'NZ_CP017176.1 Burkholderia mallei strain Bahrain1 chromosome 2, complete sequence',\n", + " 'NZ_CP016894.1 Bacillus subtilis strain HJ0-6, complete genome',\n", + " 'NZ_CP016895.1 Acinetobacter larvae strain BRTC-1 chromosome, complete genome',\n", + " 'NZ_CP016294.1 Stenotrophomonas rhizophila strain QL-P4, complete genome',\n", + " 'NZ_CP016824.1 Streptomyces sampsonii strain KJ40 chromosome, complete genome',\n", + " 'NZ_CP016341.1 Bordetella holmesii strain H903, complete genome',\n", + " 'NZ_CP016811.1 Klebsiella pneumoniae strain DHQP1002001 chromosome, complete genome',\n", + " 'NZ_CP016810.1 Klebsiella pneumoniae strain DHQP1002001 plasmid p_IncFIB_DHQP1002001, complete sequence',\n", + " 'NZ_CP016812.1 Klebsiella pneumoniae strain DHQP1002001 plasmid p_incR_DHQP1002001, complete sequence',\n", + " 'NZ_CP016342.1 Bordetella parapertussis strain H904, complete genome',\n", + " 'NZ_CP011533.1 Streptomyces noursei ATCC 11455, complete genome',\n", + " 'NZ_CP012334.1 Bordetella sp. H567, complete genome',\n", + " 'NZ_CP013749.1 Lactobacillus plantarum strain KP, complete genome',\n", + " 'NZ_CP013750.1 Lactobacillus plantarum strain KP plasmid unnamed1, complete sequence',\n", + " 'NZ_CP013751.1 Lactobacillus plantarum strain KP plasmid unnamed2, complete sequence',\n", + " 'NZ_CP013752.1 Lactobacillus plantarum strain KP plasmid unnamed3, complete sequence',\n", + " 'NZ_CP013753.1 Lactobacillus plantarum strain DF, complete genome',\n", + " 'NZ_CP013754.1 Lactobacillus plantarum strain DF plasmid unnamed1, complete sequence',\n", + " 'NZ_CP013755.1 Lactobacillus plantarum strain DF plasmid unnamed2, complete sequence',\n", + " 'NZ_CP013756.1 Lactobacillus plantarum strain DF plasmid unnamed3, complete sequence',\n", + " 'NZ_CP016431.1 Bordetella bronchiseptica strain I328 chromosome, complete genome',\n", + " 'NZ_CP029397.1 Acinetobacter defluvii strain WCHA30 chromosome, complete genome',\n", + " 'NZ_CP029389.1 Acinetobacter defluvii strain WCHA30 plasmid p1_010030, complete sequence',\n", + " 'NZ_CP029390.1 Acinetobacter defluvii strain WCHA30 plasmid p2_010030, complete sequence',\n", + " 'NZ_CP029391.1 Acinetobacter defluvii strain WCHA30 plasmid p3_010030, complete sequence',\n", + " 'NZ_CP029392.1 Acinetobacter defluvii strain WCHA30 plasmid p4_010030, complete sequence',\n", + " 'NZ_CP029393.1 Acinetobacter defluvii strain WCHA30 plasmid p5_010030, complete sequence',\n", + " 'NZ_CP029394.1 Acinetobacter defluvii strain WCHA30 plasmid p6_010030, complete sequence',\n", + " 'NZ_CP029395.1 Acinetobacter defluvii strain WCHA30 plasmid p7_010030, complete sequence',\n", + " 'NZ_CP029396.1 Acinetobacter defluvii strain WCHA30 plasmid pOXA58_010030, complete sequence',\n", + " 'NZ_CP012375.1 Microcystis aeruginosa NIES-2481, complete genome',\n", + " 'NZ_CP025929.1 Microcystis aeruginosa NIES-2481 plasmid p1, complete sequence',\n", + " ...]" + ] + }, + "execution_count": 25, + "metadata": {}, + "output_type": "execute_result" + } + ], + "source": [ + "all_fastas" + ] + }, + { + "cell_type": "code", + "execution_count": 117, + "metadata": {}, + "outputs": [ + { + "data": { + "text/html": [ + "<div>\n", + "<style scoped>\n", + " .dataframe tbody tr th:only-of-type {\n", + " vertical-align: middle;\n", + " }\n", + "\n", + " .dataframe tbody tr th {\n", + " vertical-align: top;\n", + " }\n", + "\n", + " .dataframe thead th {\n", + " text-align: right;\n", + " }\n", + "</style>\n", + "<table border=\"1\" class=\"dataframe\">\n", + " <thead>\n", + " <tr style=\"text-align: right;\">\n", + " <th></th>\n", + " <th>n_sequences</th>\n", + " <th>species</th>\n", + " </tr>\n", + " <tr>\n", + " <th>genus</th>\n", + " <th></th>\n", + " <th></th>\n", + " </tr>\n", + " </thead>\n", + " <tbody>\n", + " <tr>\n", + " <th>Bacillus</th>\n", + " <td>1132</td>\n", + " <td>11</td>\n", + " </tr>\n", + " <tr>\n", + " <th>Streptomyces</th>\n", + " <td>743</td>\n", + " <td>5</td>\n", + " </tr>\n", + " <tr>\n", + " <th>Vibrio</th>\n", + " <td>468</td>\n", + " <td>5</td>\n", + " </tr>\n", + " <tr>\n", + " <th>Rhizobium</th>\n", + " <td>325</td>\n", + " <td>6</td>\n", + " </tr>\n", + " <tr>\n", + " <th>Pseudomonas</th>\n", + " <td>304</td>\n", + " <td>8</td>\n", + " </tr>\n", + " <tr>\n", + " <th>Staphylococcus</th>\n", + " <td>301</td>\n", + " <td>6</td>\n", + " </tr>\n", + " <tr>\n", + " <th>Clostridium</th>\n", + " <td>259</td>\n", + " <td>5</td>\n", + " </tr>\n", + " <tr>\n", + " <th>Streptococcus</th>\n", + " <td>222</td>\n", + " <td>6</td>\n", + " </tr>\n", + " <tr>\n", + " <th>Planktothrix</th>\n", + " <td>179</td>\n", + " <td>5</td>\n", + " </tr>\n", + " <tr>\n", + " <th>Stenotrophomonas</th>\n", + " <td>176</td>\n", + " <td>5</td>\n", + " </tr>\n", + " <tr>\n", + " <th>Escherichia</th>\n", + " <td>133</td>\n", + " <td>3</td>\n", + " </tr>\n", + " <tr>\n", + " <th>Paenibacillus</th>\n", + " <td>127</td>\n", + " <td>3</td>\n", + " </tr>\n", + " <tr>\n", + " <th>Listeria</th>\n", + " <td>104</td>\n", + " <td>4</td>\n", + " </tr>\n", + " <tr>\n", + " <th>Corynebacterium</th>\n", + " <td>103</td>\n", + " <td>7</td>\n", + " </tr>\n", + " <tr>\n", + " <th>Klebsiella</th>\n", + " <td>16</td>\n", + " <td>3</td>\n", + " </tr>\n", + " <tr>\n", + " <th>Shigella</th>\n", + " <td>14</td>\n", + " <td>3</td>\n", + " </tr>\n", + " <tr>\n", + " <th>Salmonella</th>\n", + " <td>4</td>\n", + " <td>2</td>\n", + " </tr>\n", + " <tr>\n", + " <th>Enterobacter</th>\n", + " <td>1</td>\n", + " <td>1</td>\n", + " </tr>\n", + " </tbody>\n", + "</table>\n", + "</div>" + ], + "text/plain": [ + " n_sequences species\n", + "genus \n", + "Bacillus 1132 11\n", + "Streptomyces 743 5\n", + "Vibrio 468 5\n", + "Rhizobium 325 6\n", + "Pseudomonas 304 8\n", + "Staphylococcus 301 6\n", + "Clostridium 259 5\n", + "Streptococcus 222 6\n", + "Planktothrix 179 5\n", + "Stenotrophomonas 176 5\n", + "Escherichia 133 3\n", + "Paenibacillus 127 3\n", + "Listeria 104 4\n", + "Corynebacterium 103 7\n", + "Klebsiella 16 3\n", + "Shigella 14 3\n", + "Salmonella 4 2\n", + "Enterobacter 1 1" + ] + }, + "execution_count": 117, + "metadata": {}, + "output_type": "execute_result" + } + ], + "source": [ + "counts = inventory.reset_index().groupby(\"genus\").agg({\"count\", sum}).drop((\"species\"), axis=1)\n", + "counts.columns=[\"n_sequences\",\"species\"]\n", + "counts.sort_values(\"n_sequences\", ascending=False)" + ] + }, + { + "cell_type": "code", + "execution_count": null, + "metadata": {}, + "outputs": [], + "source": [] + } + ], + "metadata": { + "kernelspec": { + "display_name": "Python 3", + "language": "python", + "name": "python3" + }, + "language_info": { + "codemirror_mode": { + "name": "ipython", + "version": 3 + }, + "file_extension": ".py", + "mimetype": "text/x-python", + "name": "python", + "nbconvert_exporter": "python", + "pygments_lexer": "ipython3", + "version": "3.6.8" + }, + "varInspector": { + "cols": { + "lenName": 16, + "lenType": 16, + "lenVar": 40 + }, + "kernels_config": { + "python": { + "delete_cmd_postfix": "", + "delete_cmd_prefix": "del ", + "library": "var_list.py", + "varRefreshCmd": "print(var_dic_list())" + }, + "r": { + "delete_cmd_postfix": ") ", + "delete_cmd_prefix": "rm(", + "library": "var_list.r", + "varRefreshCmd": "cat(var_dic_list()) " + } + }, + "types_to_exclude": [ + "module", + "function", + "builtin_function_or_method", + "instance", + "_Feature" + ], + "window_display": false + } + }, + "nbformat": 4, + "nbformat_minor": 2 +}