TranSiGen

code for "Deep Representation Learning of Chemical-induced Transcriptional Profile for Phenotype-Based Drug Discovery"

Fig. 1 TranSiGen’s architecture and application. a The data processing flow for TranSiGen. b The architecture and inference process of TranSiGen. c The applications of TranSiGen-derived representation.

Train model

python train_TranSiGen_full_data.py --data_path ../data/LINCS2020/data_example/processed_data_id.h5
                                    --molecule_path ../data/LINCS2020/idx2smi.pickle
                                    --molecule_feature KPGT
                                    --initialization_model pretrain_shRNA
                                    --split_data_type smiles_split
                                    --n_epochs 300
                                    --n_latent 100
                                    --molecule_feature_embed_dim 400
                                    --batch_size 64
                                    --learning_rate 1e-3
                                    --beta 0.1
                                    --dropout 0.1
                                    --weight_decay 1e-5
                                    --train_flag True
                                    --eval_metric True

Infer profile

python prediction.py --model_path   ../results/trained_models_164_cell_smiles_split/364039/feature_KPGT_init_pretrain_shRNA/best_model.pt
                     --data_path ../data/PRISM/screening_compound.csv
                     --molecule_feature_embed_path ../data/PRISM/KPGT_emb2304.pickle
                     --cell YAPC
                     --seed 100

(infer_weight.gctx for inference has been uploaded in Release)

Setup and dependencies

requirements.txt contains environment of this project.

Requirements

python = 3.6.13
pytorch = 1.5.1
cmappy = 4.0.1
scikit-learn = 0.24.2
numpy = 1.19.5
rdkit = 2020.09.1

Update

The following required files can be downloaded from the Release:
processed_data.h5
infer_weight.gctx
ligand_based_virtual_screening.rar