How mozaffarilegha/drugcombinationprediction_drs works
DrugCombinationPrediction_DRS predicts oncology drug-combination synergy using resistance-signature and other biological features, with conventional ML workflows and the SynergyX deep-learning model. The principal SynergyX path loads fold data, constructs cell/drug examples, trains and evaluates the model, and reports synergy scores. The README also describes benchmark evaluation and cross-dataset validation. Some ML model implementations and SynergyX internals are indexed but not excerpted, so their details and unshown connections are kept broad.
Main entry points: My_inMemory_dataset.py Synergy Dataset · main.py Training Runner · utlis.py Training Utilities · model.py SynergyX Network
Components 19
Biological Features
- DEG Features
models/ML models/prepareLINCS.py - Sensitivity DEG
models/ML models/prepareLINCSsenDEG.py - Substructure Features
models/ML models/prepareLINCSsub.py - Drug Embeddings
models/ML models/prepareSubstructure.py
Classical ML
- ML Estimators
models/ML models
SynergyX
- Training Runner
models/SynergyX/main.py - Training Utilities
models/SynergyX/utlis.py - Synergy Dataset
models/SynergyX/dataset/My_inMemory_dataset.py - Dataset Base
models/SynergyX/dataset/base_InMemory_dataset.py - SynergyX Network
models/SynergyX/models/model.py - Attention Encoders
models/SynergyX/models/model_utlis.py - Fusion Head
models/SynergyX/models/head.py - Regression Metrics
models/SynergyX/metrics.py
Datasets
- Cell Expression Data
data/985_cellGraphs_exp_mut_cn_eff_dep_met_4079_genes_norm.npy - DEG Signature Data
data/DEGWithMASK.npy - Sensitivity Signatures
data/DEGsenWithMASK.npy - Drug Feature Data
data/drugSmile_drugSubEmbed_2644.npy - Synergy Benchmarks
dbs
Other components
- Researcher
Connections 17
- Researcher → Training Runner: runs
models/SynergyX/main.py - Training Runner → Training Utilities: loads folds
models/SynergyX/main.py - Training Utilities → Synergy Dataset: constructs datasets
models/SynergyX/utlis.py - Synergy Dataset → Dataset Base: inherits
models/SynergyX/dataset/My_inMemory_dataset.py - Synergy Dataset → Cell Expression Data: loads features
models/SynergyX/dataset/My_inMemory_dataset.py - Synergy Dataset → Drug Feature Data: loads features
models/SynergyX/dataset/My_inMemory_dataset.py - Training Runner → SynergyX Network: uses model
models/SynergyX/main.py - SynergyX Network → Attention Encoders: uses encoders
models/SynergyX/models/model.py - SynergyX Network → Fusion Head: uses head
models/SynergyX/models/model.py - Training Utilities → Regression Metrics: evaluates predictions
models/SynergyX/utlis.py - DEG Features → DEG Signature Data: uses signatures
- Sensitivity DEG → Sensitivity Signatures: uses signatures
- Substructure Features → Cell Expression Data: uses cell features
models/ML models/prepareLINCSsub.py - Drug Embeddings → Drug Feature Data: loads embeddings
models/ML models/prepareSubstructure.py - Researcher → ML Estimators: runs predictions
- Synergy Benchmarks → ML Estimators: benchmark data
- Synergy Benchmarks → Training Runner: validation data