a b/docs/Install.md 

  1
  





  
  

  2
  
# Installation Guide





  
  

  3
  





  
  

  4
  
## Prerequisites:





  
  

  5
  
 ### The required packages for running DeepMod are listed below:





  
  

  6
  
    * Python 3.6





  
  

  7
  
    * python packages:





  
  

  8
  
        + h5py





  
  

  9
  
        + numpy





  
  

  10
  
        + scipy





  
  

  11
  
    * BWA MEM/minimap2





  
  

  12
  
    * SAMTOOLS





  
  

  13
  
    * HDF5





  
  

  14
  
    * Tensorflow





  
  

  15
  
    * If running performance evaluation, you might need





  
  

  16
  
        + python packages: rpy2, matplotlib, scikit-learn





  
  

  17
  
        + R packages: ggplot2, gridExtra, plyr, RColorBrewer





  
  

  18
  





  
  

  19
  
The packages I used are 





  
  

  20
  
```





  
  

  21
  
h5py                      2.7.1





  
  

  22
  
hdf5                      1.10.1





  
  

  23
  
numpy                     1.14.0





  
  

  24
  
scikit-learn              0.19.1





  
  

  25
  
scipy                     1.0.0





  
  

  26
  
tensorflow                1.7.0





  
  

  27
  
samtools                  1.5





  
  

  28
  
minimap2                  2.12





  
  

  29
  
bwa                       0.7.15





  
  

  30
  
matplotlib                2.1.2





  
  

  31
  
rpy2                      2.8.6





  
  

  32
  
r                         3.4.2





  
  

  33
  
```





  
  

  34
  





  
  

  35
  
  ### Package installation      





  
  

  36
  
Users have their own choice of how to install required package above. But to efficiently manage the installation packages, Anaconda is recommended. After installing Annocoda, it would also be an good option to use virtual environment in annocoda. `conda create -n mdeepmod python=3.6` can be used to create a virtual environment, and `source activate mdeepmod` can be used to activate the virtual environment, and then install required packages or run DeepMod. If users want to exit the virtual environment, simply type `source deactivate`. 





  
  

  37
  





  
  

  38
  
After creating `mdeepmod` virtual environment using annocoda, the following commands can install majority of necessary packages:





  
  

  39
  





  
  

  40
  
```





  
  

  41
  
source activate mdeepmod





  
  

  42
  
conda install -c anaconda h5py hdf5 numpy scipy scikit-learn





  
  

  43
  
conda install -c bioconda minimap2 bwa samtools





  
  

  44
  
conda install -c r rpy2 r-ggplot2 r-gridextra r-plyr r-rcolorbrewer





  
  

  45
  
```





  
  

  46
  
Users might need to install `tensorflow` with version 1.13 by yourself or by using `conda install -c conda-forge tensorflow matplotlib` (if there are conflicts when using this command, uers need to give up and install `tensorflow` and `matplotlib` by yourself.).





  
  

  47
  





  
  

  48
  
### Additional notes





  
  

  49
  





  
  

  50
  
Some users reported that the compression format of their fast5 is vbz instead of gzip. Install `ont-vbz-hdf-plugin` solves the problem. Otherwise, an error message such as "Error!!! No Raw_reads/Signal data" will be shown.





  
  

  51
  





  
  

  52
  





  
  

  53
  
## Install DeepMod:





  
  

  54
  
    * git clone https://github.com/WGLab/DeepMod





  
  

  55
  
    * The script to be run is in bin/DeepMod.py: 





  
  

  56
  
        + Run `python bin/DeepMod.py` for help document or 





  
  

  57
  
        + Run `python bin/DeepMod.py detect` for the detection help document.





  
  

  58
  
Users might need to run `source activate mdeepmod` if the virtual environment and required packages are installed with the commands above.





  
  

  59
  





  
  

  60
  
## Installation time





  
  

  61
  
Without GPU-version tensorflow, it would take ~30 minutes to install required packages and DeepMod. 





  
  

  62
  





  
  

  63
  
## Usage:





  
  

  64
  
 For how to use them, please refer to [Usage](https://github.com/WGLab/DeepMod/blob/master/docs/Usage.md)





  
  

  65