Benchmarking tissue- and cell type-of-origin deconvolution in cell-free transcriptomics: analysis code

Le résumé fourni par la source

This software archive contains the custom code supporting the study “Benchmarking tissue- and cell type-of-origin deconvolution in cell-free transcriptomics.” The repository provides workflows for constructing multi-organ tissue- and cell-type reference profiles, generating simulated tissue mixtures and single-cell-derived pseudobulk mixtures, running conventional and deep-learning deconvolution approaches, evaluating deconvolution accuracy and robustness, analysing published plasma cell-free RNA cohorts, and reproducing the principal and supplementary benchmarking results. The analyses include tissue-of-origin and cell type-of-origin benchmarking across alternative reference configurations; perturbation analyses modelling negative-binomial noise, transcript degradation, and reduced gene detectability; evaluation using absolute error, mean absolute error, Pearson correlation, and Jensen–Shannon divergence; reference-signature similarity analyses; and biological validation using published clinical cfRNA cohorts. Input sequencing datasets are not redistributed in this archive. Public data accessions, preprocessing instructions, software versions, and parameter configurations required to reproduce the analyses are provided in the repository documentation.

Ce résumé expose les affirmations des auteurs. BNTIC ne l’interprète pas comme une validation indépendante des résultats.