RMA-normalized rat-liver Affymetrix toxicogenomics matrices (4 GEO Series) for DILIrank 2.0 prediction

BrainArray Entrez-CDF RMA-normalized rat liver gene expression matrices from four GEO Series (GSE57805, GSE119933, GSE68110, GSE40117) — 2,146 samples (in vivo + in vitro hepatocyte) across ~240 unique drug perturbations, processed for integration with the merged-rat-liver-DILIrank prediction pipeline. Pipeline / processing scripts: github.com/Ajay1989Kumar/merged-rat-liver-dili-prediction Source GEO Series: GSE57805 (DrugMatrix rat hepatocyte, Affymetrix Rat230_2, 939 samples, 237 compounds); GSE119933 (TGP+NRW rat hepatocyte directory, Rat230_2, 547 samples); GSE68110 (short-term carcinogens/non-carcinogens panel, Rat Expression 230A, 421 samples); GSE40117 (rat-liver in vitro hepatocarcinogen panel, Rat230_2, 239 rat samples). Processing: oligo::rma() with BrainArray v25.0.0 Entrez-CDF packages (rat2302rnentrezgcdf, rae230arnentrezgcdf). GSE57805 was processed in 4 batches of ~234 samples due to preprocessCore memory limits at higher sample counts; two corrupt/truncated CEL files were excluded (GSM1389850, GSM1390552). Files: rma_*_brainarray.tsv — gene (Entrez ID) x sample matrix, log2 RMA intensities; all_samples_metadata.tsv — per-sample annotations (GSM, title, platform, organism, is_affy_rat, is_liver, is_control, compound, dose, time_h, cel_file).