etiennereboul / larvae

VAE generating drug like ligand with high affinity to protein or RNA target

LaRVAE

i) Installation Your first time using LaRVAE do the following:

1. Log into compute canada cluster

2. Load the correct modules with the command
   $ module load gcc rdkit/2021.09.3 openbabel/3.1.1 python/3.8

3. Save the modules in a collection
   $ module save larvaeModules

4. Clone the git repository into your working directory with the command
   $ git clone https://github.com/EtienneReboul/LaRVAE.git

5. After cloning the LaRVAE repository you should set up a virtual environment using the following commands:
   $ python3 -m pip install --user virtualenv
   $ python3 -m venv /pathToNewEnv/EnvName
   $ source /pathToNewEnv/EnvName/bin/active

6. Then install the packages with the command
   $ pip install -r /pathToLaRVAE/requirements_3.8.10.txt

7. To get data run the following scripts $ python3 scripts1/download_moses.py
   $ python3 scripts1/build_vocab.py

8. Once you are done running LaRVAE deactivate the virtual environment with the command
   $ deactivate

Subsequent times using LaRVAE do the following:

1. Restore the module collection
   $ module restore larvaeModules

2. Activate the virtual environment
   $ source /pathToNewEnv/EnvName/bin/active

3. Once you are done running LaRVAE deactivate the virtual environment with the command
   $ deactivate

ii) Download Data

1. Download smile data from moses
   $ python3 scripts1/download_moses.py

2. Covert smile data into seflie data
   $ python3 scripts1/get_selfies.py

iii) Train Initial Model
1a) Use tranvae to train an initial vae: $python scripts1/train.py --model rnnattn --data_source custom --train_mols_path my_train_data.txt --test_mols_path my_test_data.txt --vocab_path my_vocab.pkl --char_weights_path my_char_weights.npy --save_name my_model
1b) if you want to train using SELFIE adjacency matrices you can use the option "--adj_matrix True" and you can set the weight for non-edges of the adjacency matrix with the option "--adj_weight " followed by a float in (0,1)

2. You can use "$ scripts1/train.py --help" to see all command options

iv) Launch CBAS (note: conditioning by adaptive sampling is not implemented in the current version of LaRVAE)

1. Launch slurm_master.py for 3 iterations with a command of the form $salloc --time=0:30:0 --ntasks=1 --cpus-per-task=10 --mem-per-cpu=2048M --account=def-jeromew srun python cbas/slurm_master.py --name launchName --iters 3

2. Look in scripts1/parsers1.py to see all options for the sampler, docker and train scripts