Analysis code and supporting data for "IOBRpy enables agentic multi-omics decoding of anti-tumor immunity"

This record contains the manuscript-specific analysis code and supporting derived data for the Nature Protocols study “IOBRpy enables agentic multi-omics decoding of anti-tumor immunity”. The deposited materials support the Python–R benchmarking and the agent-skill-guided multidimensional output demonstration presented in Figures 2 and 4. They include analysis and visualization scripts, software environment information, source tables and supporting derived results. The analyses were performed using IOBRpy v0.2.0. Figure 2 evaluates runtime behaviour and Python–R output concordance for representative tumour-microenvironment profiling methods using a 936-sample OAK/POPLAR (OP) expression matrix. Figure 4 uses a ten-sample subset comprising five OP immunotherapy (OP-IO) and five OP chemotherapy (OP-Chemo) samples to demonstrate agent-guided review of immune-deconvolution outputs, signature scores, ligand-receptor scores, HLA typing results and TCR repertoire metrics. The original OAK/POPLAR sequencing and clinical data are controlled-access datasets. The original data are available through the European Genome-phenome Archive under the following study and dataset accessions: OAK and POPLAR:EGAS00001005013EGAD00001007703 Please refer to README.md for file descriptions, software requirements, input-data requirements and reproduction instructions.