Node-Sparing Radiotherapy Boosts ICB Efficacy in pMMR Rectal Cancer via TDLN-Tumor CD8⁺ T Cell Clonal Expansion

Files Included File Size Description phaseII_landscape.h5ad 6.60 GB AnnData object from 3′ whole-transcriptome scRNA-seq (all cell types); contains normalized expression matrix, UMAP coordinates, and full cell metadata. phaseii_cd8_upload.h5ad 1.14 GB AnnData object from 5′ scRNA-seq + TCR-seq, CD8⁺ T cells only; includes treatment-naive (PRE) and post-treatment (POST) samples; contains TCR clonotype annotations, subtype labels (leiden2CD8celltype2), and functional signature scores. phaseii_cd8_merge_mid_upload.h5ad 1.31 GB Extended CD8⁺ T cell AnnData built on phaseii_cd8_upload.h5ad with mid-treatment samples additionally merged in; provided for analyses spanning all three treatment timepoints. fig2_landscape.ipynb 4.36 MB Jupyter notebook: 3′ whole-cell scRNA-seq data processing and analysis (corresponds to Figure 2). CRT_CD8_unified_analysis_executed.ipynb 21.66 MB Jupyter notebook: 5′ CD8⁺ T cell scRNA-seq and TCR repertoire unified analysis (corresponds to Figure 3-6). standard.h5ad 166 MB AnnData object from 5′ scRNA-seq + TCR-seq, CD8⁺ T cells only from paired tumor and lymph node samples across 6 patients with "standard" regimen. adata.X – Normalized, log1p-transformed gene expression matrix (float32, sparse). Cells were normalized using scran normalization followed by log1p transformation. Row index (adata.obs.index): cell barcodes in the format {barcode}_{sample_name}. Column index (adata.var.index): HGNC gene symbols (19,588 genes). adata.layers["counts"] – Raw UMI count matrix (integer-valued, float32, sparse) prior to normalization. Recommended for differential expression analysis or re-normalization workflows. adata.obsm["X_gex"] – 10-dimensional GEX latent embedding (float32, dense; 50,633 × 10) used for neighborhood graph construction and clustering. adata.obsm["X_umap"] – 2D UMAP coordinates (float32, dense; 50,633 × 2) computed from X_gex for visualization. adata.obs contains the following per-cell metadata: tcr: Unique TCR identifier constructed by concatenating IR_VJ_1_junction_aa, IR_VDJ_1_junction_aa, IR_VJ_1_v_call, IR_VJ_1_j_call, IR_VDJ_1_v_call, IR_VDJ_1_j_call, and individual (hyphen-delimited). Cells without a paired TCR are NaN. IR_VJ_1_junction_aa: TCR α-chain CDR3 amino acid sequence (IMGT junction definition). IR_VDJ_1_junction_aa: TCR β-chain CDR3 amino acid sequence (IMGT junction definition). IR_VJ_1_junction: TCR α-chain CDR3 nucleotide sequence. IR_VDJ_1_junction: TCR β-chain CDR3 nucleotide sequence. IR_VJ_1_v_call / IR_VJ_1_j_call: TCR α-chain V and J gene segments (IMGT nomenclature). IR_VDJ_1_v_call / IR_VDJ_1_j_call: TCR β-chain V and J gene segments (IMGT nomenclature). sample_name: Sample identifier (format: {PatientID}-{treat}-{tissue}, e.g. "31-POST-TUMOR"). individual: Patient ID used in TCR construction (POST-#1 to POST-#8; PRE-#1 to PRE-#4). treat: Treatment timepoint. PRE: pre-CRT biopsy; POST: post-CRT surgical specimen. response: Pathological response to chemoradiotherapy. pCR: complete response (ypT0N0); non-pCR: residual disease; PRE: pre-treatment (no response assigned). tissue: Tissue compartment of origin (Tumor / LymphNode). leiden2CD8celltype2: CD8⁺ T cell subtype annotation. Tn: naïve; Tpex: precursor exhausted; Trm-like: tissue-resident memory-like; Tem.str: effector memory stem-like; Tex.trans1/Tex.trans2: transitional exhausted; Tex.eff: effector-like exhausted; Tex.str: stem-like exhausted; Tex.term/IEL: terminally exhausted/intraepithelial lymphocyte-like. Naïve, Cytotoxicity, Exhaustion, TCR Signaling, Cytokine/Cytokine receptor, Chemokine/Chemokine receptor, IFN Response, Anergy, Stress response: Per-cell functional gene signature scores (float64, z-scored across all cells).