Benchmark datasets for Multi-objective domain adaptation for pan-cancer transfer of pharmacogenomic drug response to patient tumours

This is a companion repository to updated version of the Multi-Objective Deconfounding Autoencoder (https://github.com/UEFBiomedicalInformaticsLab/MODAEplus) and associated publication: "Multi-objective domain adaptation for pan-cancer transfer of pharmacogenomic drug response to patient tumours". This repository contains datasets obtained from various sources that are subject to separate Terms of Use and are included here solely to support the reproducibility of the associated research publication. The datasets include: Pre-processed gene-expression data from patients and cell-lines in TCGA (retrieved using curatedTCGAData), and CCLE (retrieved from DepMap data portal). Curated TCGA survival data from Liu et al. (2018). Pre-processed cell-line drug-sensitivity data and other files from Xia et al. (2022) supplement. Alternative cell-line drug-sensitivity data from CCLE and CTRP retrieved using PharmacoGx. Drug target associations from the Comparative Toxicogenomics Database, PharmGKB, and OpenTargets. A copy of corrected TCGA gene-expression from the Celligner (Warren et al. 2021) supplement. A copy of TCGA drug information pre-processed by He et al. (2022) together with CODE-AE embeddings, predictions, and other files in the supplement to their publication.