生信 Workflow 范畴路由

当用户目标是从生物数据到结果表、图、报告或可运行流程时，使用本范畴。先选择具体 workflow skill，不要停留在本层。

具体 skill 路由

• bulk RNA-seq、count matrix、差异表达、富集前置：读取 ./rnaseq-differential-expression/SKILL.md
• 10x/scRNA-seq、QC、聚类、marker、注释、scVI：读取 ./single-cell-rna-analysis/SKILL.md
• 空间转录组、CITE-seq、multiome、跨模态整合：读取 ./spatial-multiomics-analysis/SKILL.md
• germline/somatic variant、CNV/SV、VCF 注释：读取 ./variant-calling-interpretation/SKILL.md
• ATAC-seq、ChIP-seq、methylation、Hi-C、GRN：读取 ./epigenomics-regulation/SKILL.md
• microbiome、metagenomics、eDNA、pathogen genomics：读取 ./microbiome-metagenomics/SKILL.md
• proteomics、metabolomics、lipidomics、mzML：读取 ./proteomics-metabolomics/SKILL.md
• pathway、multi-omics、metabolic model、systems biology：读取 ./multiomics-systems-biology/SKILL.md
• TCR/BCR、neoantigen、MHC/epitope：读取 ./immune-repertoire-neoantigen/SKILL.md
• 蛋白设计流水线、RFdiffusion -> ProteinMPNN -> SimpleFold/Protenix/AlphaFold3 结构验证、候选排序：读取 ./protein-design-structure-validation/SKILL.md
• genome assembly、annotation、comparative genomics：读取 ./genome-assembly-annotation/SKILL.md
• Nanopore/PacBio、long-read QC、SV、Iso-Seq、direct RNA、nanopore methylation、phasing：读取 ./long-read-sequencing-analysis/SKILL.md
• alternative splicing、Ribo-seq、small RNA、CLIP-seq、m6A/epitranscriptomics、RNA modification：读取 ./post-transcriptional-regulation/SKILL.md
• CRISPR screen、Perturb-seq、DepMap/gene essentiality、pooled screen QC、MAGeCK/PinAPL-Py：读取 ./functional-genomics-screens/SKILL.md

交接规则

输出时至少整理：

• assay_or_pipeline
• input_object
• organism_or_taxon
• reference_build_or_database
• sample_metadata
• workflow_stages
• qc_checkpoints
• expected_outputs
• execution_entry

如果需要写脚本或配置，交给 onescience-skill -> onescience-coder。如果需要提交运行，追加 platform-engineer 并交给 onescience-runtime。