Loss of Vpr-driven TRAIL-R2 expression protects HIV-infected cells from non-cannonical NK cell TRAIL attack

This record accompanies Grasberger et al., "Loss of Vpr-driven TRAIL-R2 expression protects HIV-infected cells from non-cannonical NK cell TRAIL attack." It archives the source code and processed single-cell data underlying the secondary analysis of the Wei et al. (Immunity 2023; GEO: GSE239909) CITE-seq dataset presented in the paper. Study context The associated manuscript defines transcriptional and surface-protein signatures of HIV-infected CD4+ T cells that survive natural killer (NK) cell attack in vitro. To examine clinical relevance, a bulk RNA-seq-derived gene signature of those in vitro survivors was projected onto the Wei et al. CITE-seq atlas of HIV RNA+ cells from people with HIV. This record contains the pipeline used to perform that projection, together with the intermediate and final objects it produces. Pipeline An ordered seven-step R workflow: Per-sample 10x matrices are filtered, merged, normalized, integrated with Harmony, and embedded with UMAP. Four canonical transcriptional macro-programs (Proliferation, Cytotoxic, AP-1/TNF, IRF/IFN) are scored and assigned per cluster. HIV RNA+ annotations (bc_RNA_HIV.txt) are imported, with RNA+ calls enforced at HIV copies ≥ 2. Two gene signatures are scored on every cell: the bulk RNA-seq-derived profile of NK-cell survivors from this study (our profile; Clayton DEG list) and the Wei et al. article-derived HIV RNA+ profile. Each signature is split into UP/DOWN sets, module-scored, combined into a signed score, and z-standardized. Profile-positive calls are made at z ≥ 1.0, 1.5, and 2.0. Per-macro prevalence tables and UMAP triptychs are produced at each threshold. Concordance between the two profiles is quantified by accuracy, Cohen's κ, MCC, Jaccard, and F1, with supporting scatter, Bland-Altman, Venn, and bar plots. Composition and correlation panels are computed within the HIV RNA+ subset (Bulk+ / Wei+ / Both / Neither). Contents GSE239909_RAW/ — thirty 10x files (matrix, features, barcodes) from GEO GSE239909, mirrored here for long-term reproducibility. bc_RNA_HIV.txt — per-cell HIV RNA+ call table (cell, HIV_RNA, Copies). Clayton_DEG_List.xlsx — bulk RNA-seq DEG list used to build the our profile. Article_DEG_list.xlsx — Wei et al.-derived DEG list used to build the article profile. Outputs_Code1_20251130/ through Outputs_Code7_20251130/ — per-step Seurat RDS objects, metadata tables, prevalence tables, concordance metrics, and figure panels. MANIFEST.sha256, README.md, LICENSE, CITATION.cff — integrity, documentation, license, and citation metadata. Running the pipeline The R scripts and Dockerfile are in the companion GitHub repository. After downloading this record, point the pipeline at the extracted directory and run scripts/run_pipeline.sh (or the provided Docker image). Full instructions are in the README. Licensing Source code is released under the MIT License. Data artifacts are released under Creative Commons Attribution 4.0 International (CC-BY-4.0). Raw sequencing data originate from GEO GSE239909 and are redistributed here unmodified; please cite the Wei et al. (2023) publication alongside this record.