LUAD immune phenotyping

Code and manuscript source for a retrospective analysis of immune phenotypes and overall survival in TCGA lung adenocarcinoma (TCGA-LUAD) bulk RNA-seq profiles.

The workflow integrates transcriptome deconvolution features, engineered immune-landscape metrics, phenotype clustering, supervised reproduction of phenotype assignments, survival analysis, and an exploratory immune-cell interaction network score. The reported associations are retrospective and internally evaluated; they are not a clinical decision tool.

Repository layout

• paper/: self-contained LaTeX manuscript, bibliography, figures, and final PDF
• results/: processed result summaries and publication figures
• pipeline_core.py, novel_*.py, survival_analysis.py: principal analysis modules
• requirements.txt: Python dependencies
• Run_GEMDeCan_*.R: R entry points used for deconvolution

Raw TCGA data, local environments, installed R libraries, and trained model binaries are intentionally excluded. TCGA-LUAD source data are available from the NCI Genomic Data Commons.

Reproducibility

The repository preserves the code and processed outputs used for the manuscript. Re-running the complete workflow requires obtaining the source data described in the manuscript and installing both Python and R dependencies. No new experiments were run during preparation of this archival release.

The final manuscript is paper/luad_immune_phenotyping_preprint.pdf. Compile the source from paper/ with latexmk -pdf main.tex. The published preprint is permanently available from Zenodo.

Citation

Vetsa, J., & Veerapaneni, S. V. (2026). Transcriptome-Based Immune Phenotyping and Survival Stratification in Lung Adenocarcinoma (Version 1). Zenodo. https://doi.org/10.5281/zenodo.21630124

BibTeX:

@misc{vetsa2026luad,
  author    = {Vetsa, Jayaditya and Veerapaneni, Sai Venkat},
  title     = {Transcriptome-Based Immune Phenotyping and Survival Stratification in Lung Adenocarcinoma},
  year      = {2026},
  publisher = {Zenodo},
  version   = {1},
  doi       = {10.5281/zenodo.21630124},
  url       = {https://doi.org/10.5281/zenodo.21630124}
}

Code is released under the MIT License. The manuscript text and figures are released under CC BY 4.0.