Installing tensorqtl module

Installing tensorqtl requires pytorch which is based on gpus but there is also a cpu based version.

CRI has set up pytorch for cpus as in a conda environment and that is what I am going to use to set up tensorqtl.

I will install the tensorqtl in im-lab share space for lab use.

Steps for installation

Create a directory for the environment

mkdir -p /gpfs/data/im-lab/nas40t2/bin/envs
cd /gpfs/data/im-lab/nas40t2/bin/envs

Copy the pytorch environment into this new directory and name it tensorqtl
```
cp -r /apps/software/gcc-6.2.0/miniconda3/4.7.10/envs/pytorch-1.4.0-cpu_py37 tensorqtl
```
Now we have the pytorch setup environment next we are going to set up tensorqtl
Checking if requirements are available

Activate the conda environment
```
conda activate /gpfs/data/im-lab/nas40t2/bin/envs/tensorqtl
```
In this environment when you test the pip command its not executable because the environment has python2. We need to upgrade the environment to use python3.
Test the availability of pip and python3 using the following commands
```
python3 --version
pip3 --version
pip --version
```
If you get error then you definitely need to set up these tools
Set up python3

Install python3 which works with the set up pip
```
conda install python==3.8.0
```
Install tensorqtl

Tensorqtl is available from pip
```
pip install tensorqtl
```
Once installation is successful install the dependecies
Install the rpy2 dependency
```
conda install rpy2
```
Test tensorqtl
```
 python3 -m tensorqtl --help
```
Clean up

Conda caches all these packages which consume a lot of disk space. The need to be removed;
```
 conda clean --all
```

NB: This environment is available for lab use. To activate the the environment for use just activate it

  conda activate /gpfs/data/im-lab/nas40t2/bin/envs/tensorqtl

Happy QTL mapping!!!