bioRxivJClub
Integrative Single-Cell Transcriptomics and Explainable AI for Disease Mechanism and Biomarker Discovery in Triple-Negative Breast Cancer
bioRxiv · · Preprint · Open access
Huda, BISWAS
Abstract ↗AI summary
The abstract is read at the publisher; the summary is JClub's.
In MDA-MB-231 transwell assays, invasive cells are distinguished from non-invasive cells by a reproducible transcriptional state, enabling an XGBoost AUC of 0.92.
- Why it matters: Invasion is an early metastatic step in triple-negative breast cancer, but the transcriptional features separating invasive from non-invasive cells within the same tumour or line remain unclear.
- What they did: The authors re-analysed GSE329526 using cell-level Wilcoxon, XGBoost with random cell splits, SHAP, and Enrichr, then repeated at sample level with whole-sample cross-validation and sample-mean differential expression.
- The result: Across held-out samples, 6,651 of 11,727 genes differed (FDR < 0.05), SHAP highlighted inflammatory/stress genes, and GSEA showed higher KRAS (NES 2.30, FDR 0.001) and E2F targets (NES 1.62, FDR 0.038) with lower interferon responses.
The findingWhy it mattersWhat they didThe result
- Open access