ppromprasert/cms4-singlecell-immune-states-ancestry

CMS4 CRC single-cell immune-state analysis integrating Seurat, Harmony, Azimuth, pseudobulk DESeq2, pathway GSEA, and African ancestry. # CMS4 CRC Single-Cell Immune States and African-Ancestry Analysis **Single-cell immune-state analysis of CMS4 colorectal cancer integrating lineage-specific Seurat workflows, compartment-specific QC, Harmony batch integration, Azimuth reference annotation, patient-level subtype composition, pseudobulk DESeq2, and GO/Hallmark pathway enrichment across African-ancestry groups.** ## Scientific question Within CMS4 CRC, do tumor-infiltrating immune populations differ in **cell state, subtype composition, transcriptional programs, and pathway activity** across African-ancestry groups? This is distinct from both the cellular-composition project and the inferCNV project: ```text single-cell CRC program | +-- cellular composition repo | patient-level proportions + CMS composition | +-- THIS REPO | CD8 / CD4 / B / M2 / mast / progenitor immune states | Harmony + Azimuth + pseudobulk DE + GSEA | +-- CMS4 inferCNV repo malignant-cell CNV + clonal evolution ``` ## 1. Immune-compartment workflow Dedicated source scripts exist for **CD8 T cells, CD4 T cells, B cells, M2/TAM-like macrophages, mast cells, and CLP/progenitor-like populations**. The repeated source architecture is technically substantial. For example, the CD8 workflow reads 10x H5 matrices from four batches, merges Seurat objects, identifies CD8 candidates with a CD3D/CD3E/CD8A/CD8B module score, and then performs compartment-specific QC. The source QC thresholds are: ```text nFeature_RNA > 500 nFeature_RNA violin = ancestry-group distribution; dots = individual patients colored by KRAS status. This genotype overlay is descriptive unless formally entered into an adjusted statistical model. ## Statistical design | Layer | Biological replicate | Method | |---|---|---| | Immune-state embedding | cell | Seurat / Harmony | | Reference annotation | cell | Azimuth | | Functional markers | cell | FeaturePlot / module interpretation | | Subtype composition | patient | patient-level percentages | …