MTLSynergy

Requirements

• python=3.7
• cuda=10.2
• pytorch=1.8.1
• sklearn=1.0.2
• pandas=1.3.5

Start

Run the AEtrain.py first to pre-train a drug encoder and a cell line encoder, and then run the MTLSynergytrain.py to train the model.

Data

drugs.csv: Information of 3118 drugs.

cell_lines.csv: Information of 175 cell lines.

drug_features.csv: Features of 3118 drugs, 1213-dimensional vector for each drug.

cell_line_features.csv: Features of 175 cell lines, 5000-dimensional vector for each cell lines.

oneil_summary_idx.csv: 22 737 samples from O'Neil，each sample consists of two drugs id, a cell line id, synergy score of the drug combination on the cell line, respective sensitivity scores of the two drugs on the cell line.

Training files

AEtrain.py: used to pre-train a drug encoder and a cell line encoder.

MTLSynergytrain.py: used to train MTLSynergy in the Leave Drug Combinations Out scenario.

MTLSynergy_LeaveCellOut.py: used to train MTLSynergy in the Leave Cell Lines Out scenario.

MTLSynergy_LeaveDrugOut.py: used to train MTLSynergy in the Leave Drugs Out scenario.

GBMtrain.py: used to train Gradient Boosting Machine in the Leave Drug Combinations Out scenario.

GBM_LeaveCellOut.py: used to train Gradient Boosting Machine in the Leave Cell Lines Out scenario.

GBM_LeaveDrugOut.py: used to train Gradient Boosting Machine in the Leave Drugs Out scenario.

RFtrain.py: used to train Random Forest in the Leave Drug Combinations Out scenario.

RF_LeaveCellOut.py: used to train Random Forest in the Leave Cell Lines Out scenario.

RF_LeaveDrugOut.py: used to train Random Forest in the Leave Drugs Out scenario.

Source code of the comparative methods

PRODeepSyn: https://github.com/TOJSSE-iData/PRODeepSyn

TranSynergy: https://github.com/qiaoliuhub/drug_combination

AuDnnSynergy: The authors did not provide the source code.

DeepSynergy: https://github.com/KristinaPreuer/DeepSynergy