|
a |
|
b/openomics/__init__.py |
|
|
1 |
from __future__ import print_function, division, absolute_import |
|
|
2 |
|
|
|
3 |
import json |
|
|
4 |
import logging |
|
|
5 |
import os |
|
|
6 |
import sys |
|
|
7 |
from os.path import join |
|
|
8 |
|
|
|
9 |
import astropy |
|
|
10 |
import dask.dataframe as dd |
|
|
11 |
import pandas as pd |
|
|
12 |
|
|
|
13 |
"""Top-level package for openomics.""" |
|
|
14 |
|
|
|
15 |
__author__ = """Nhat (Jonny) Tran""" |
|
|
16 |
__email__ = 'nhat.tran@mavs.uta.edu' |
|
|
17 |
__version__ = '0.8.9' |
|
|
18 |
|
|
|
19 |
|
|
|
20 |
# Initialize configurations |
|
|
21 |
this = sys.modules[__name__] |
|
|
22 |
this.config = {} |
|
|
23 |
|
|
|
24 |
# Set pandas backend |
|
|
25 |
this.config["backend"] = pd |
|
|
26 |
|
|
|
27 |
# Set cache download directory |
|
|
28 |
this.config["cache_dir"] = astropy.config.get_cache_dir(this.__name__) |
|
|
29 |
logging.info("Cache directory is", this.config["cache_dir"]) |
|
|
30 |
|
|
|
31 |
home_dir = os.path.expanduser('~') |
|
|
32 |
user_conf_path = join(home_dir, ".openomics/conf.json") |
|
|
33 |
|
|
|
34 |
# Initialize user configuration file at ~/.openomics/conf.json |
|
|
35 |
if not os.path.exists(user_conf_path): |
|
|
36 |
if not os.path.exists(join(home_dir, ".openomics")): |
|
|
37 |
os.makedirs(join(home_dir, ".openomics")) |
|
|
38 |
|
|
|
39 |
if not os.path.isfile(user_conf_path): |
|
|
40 |
base_conf = {} |
|
|
41 |
base_conf['cache_dir'] = astropy.config.get_cache_dir(this.__name__) |
|
|
42 |
|
|
|
43 |
with open(user_conf_path, 'w', encoding='utf-8') as file: |
|
|
44 |
json.dump(base_conf, fp=file, indent=4) |
|
|
45 |
|
|
|
46 |
# Read configuration from ~/.openomics/conf.json |
|
|
47 |
if os.path.isfile(user_conf_path): |
|
|
48 |
try: |
|
|
49 |
with open(user_conf_path, 'r', encoding='utf-8') as file: |
|
|
50 |
user_conf = json.load(fp=file) |
|
|
51 |
|
|
|
52 |
if user_conf: |
|
|
53 |
for p in user_conf.get('database', []): |
|
|
54 |
this.config.update(p) |
|
|
55 |
|
|
|
56 |
except Exception as e: |
|
|
57 |
logging.info("Could not import configurations from", user_conf_path) |
|
|
58 |
|
|
|
59 |
# Import submodules |
|
|
60 |
|
|
|
61 |
from .transcriptomics import * |
|
|
62 |
from .genomics import * |
|
|
63 |
from .proteomics import * |
|
|
64 |
from .clinical import * |
|
|
65 |
from .multiomics import * |
|
|
66 |
|
|
|
67 |
|
|
|
68 |
def set_backend(new: str = "pandas"): |
|
|
69 |
"""Set the dataframe processing backend to either Pandas or Dask. |
|
|
70 |
|
|
|
71 |
Args: |
|
|
72 |
new (str): Either "dask" or "pandas". Default "pandas. |
|
|
73 |
""" |
|
|
74 |
assert new in ["dask", "pandas", "modin"] |
|
|
75 |
|
|
|
76 |
if new == "dask": |
|
|
77 |
this.config["backend"] = dd |
|
|
78 |
else: |
|
|
79 |
this.config["backend"] = pd |
|
|
80 |
|
|
|
81 |
|
|
|
82 |
def set_cache_dir(path: str, delete_temp: bool = False): |
|
|
83 |
"""Set the path where external databases downloaded from URL are saved. |
|
|
84 |
|
|
|
85 |
Args: |
|
|
86 |
path (str): |
|
|
87 |
delete_temp (bool): |
|
|
88 |
""" |
|
|
89 |
if not os.path.exists(path): |
|
|
90 |
raise NotADirectoryError(path) |
|
|
91 |
|
|
|
92 |
this.config["cache_dir"] = path |
|
|
93 |
astropy.config.set_temp_cache(path=path, delete=delete_temp) |
|
|
94 |
logging.info("Cache directory is", this.config["cache_dir"]) |