DGDRP

RECOVER

Exploring the space of drug combinations to discover synergistic drugs using Active Learning

TGSA

PyTorch implementation of "TGSA: Protein-Protein Association-Based Twin Graph Neural Networks for Drug Response Prediction with Similarity Augmentation"

MOFGCN

MOFGCN is a computational model designed to predict drug response in cancer cell lines by integrating multiple types of biological data. It addresses the challenge of personalized cancer treatment by leveraging multi-omics information and graph-based learning to improve prediction accuracy. ## Key Features - **Multi-omics Fusion** — Combines gene expression, copy number variation, and somatic mutation data to create a comprehensive cell line similarity network. - **Graph Convolutional Networks** — Utilizes GCNs to capture complex relationships between cell lines and drugs within a heterogeneous network. - **Multiple Experimental Modules** — Supports various validation scenarios including random clearing, single row/column clearing, single drug prediction, and targeted drug experiments. - **Performance Metrics** — Includes tools for calculating AUC, ACC, F1 score, and MCC to evaluate model predictions. - **Preprocessed Datasets** — Works with standardized data from CCLE and GDSC databases, including cell line-drug association matrices and drug fingerprint features. ## Philosophy MOFGCN is built on the principle that integrating diverse biological data types through graph neural networks can more accurately model the complex mechanisms of drug response in cancer.

DeepAEG

DeepAEG: A model for predicting cancer drug response based on data enhancement and edge-collaborative update strategies