fgcz-skills

Single-cell and spatial transcriptomics: Seurat v5, CellRanger, cytotrace2, scRepertoire, RCTD spatial deconvolution, SPLIT purification, SCEVAN CNV, SComatic somatic mutations, CellChat cell-cell communication, Slingshot trajectory/pseudotime, Xenium-to-Allen-CCF spatial registration, clusterProfiler pathway enrichment, and InSituCNV spatial CNV inference.

Deep learning and GPU-accelerated single-cell analysis: scVI/scANVI/totalVI/MultiVI batch correction, RAPIDS GPU preprocessing for large datasets (>500K cells), CyteType AI cell annotation, and rctd-py GPU-accelerated Python RCTD spatial deconvolution.

Sequencing data processing: nf-core/Nextflow pipelines, MiXCR bulk TCR/BCR repertoire, genome reference building, Draugr demultiplexing, Sample2Barcode generation, and SpaceRanger Visium/VisiumHD/CytAssist spatial transcriptomics.

B-Fabric LIMS integration: dataset registration and resource management via btools, project/sample/workunit querying, lineage tracing, and metadata export.

FGCZ infrastructure and deployment: SUSHI job submission and app development with ezRun, Lmod module management with pixi project environment setup, GitLab CI/CD on gitlab.bfabric.org, ShinyProxy app deployment, ScMultiOmics SUSHI app, promotion of one-off Rmd analyses into SUSHI-shaped artefacts for B-Fabric/SUSHI lineage, and autonomous self-correcting R Markdown / Python report rendering.

LC-MS untargeted metabolomics feature curation with automated QC metrics and interactive HTML reports.

Meta-skills for the fgcz-skills marketplace: shared FGCZ environment context, multi-LLM review workflow, and self-skill auditing. Load when working on any FGCZ task that benefits from institutional context, or when authoring/reviewing skills in this marketplace.