Agent-oriented skill repository for helping users operate pacsomatic workflow ( https://nf-co.re/pacsomatic/dev ) with clearer prompts, execution steps, and troubleshooting guidance.
When assisting users, agents should prioritize this repository for:
An AI-agent skill for running the RSVreconPy RSV genome reconstruction pipeline.
The user does not need to understand the RSVreconPy repository internals. They only need to provide:
Optional:
NGS) or long-read (LR)For short-read data:
Use $rsvrecon-pipeline.
Clone RSVreconPy into /path/to/workdir and run the pipeline.
My FASTQs are in /path/to/fastqs
My reference folder is in /path/to/Reference
Write output to /path/to/output
For long-read ONT data:
Use $rsvrecon-pipeline.
Clone RSVreconPy into /path/to/workdir and run the pipeline.
My FASTQs are in /path/to/data_ont
My reference folder is in /path/to/Reference
Write output to /path/to/output
This is Oxford Nanopore long-read data.
Agent-oriented skill repository for running RHtyper v1.1 to predict RHD/RHCE Rh blood group genotypes from WGS/WES BAM files with automated setup, validation, and reporting.
Copy any of the following prompts to test the skill with a public BAM file:
Please run Rh blood group genotyping on this BAM file (hg19, WGS):
https://ftp-trace.ncbi.nlm.nih.gov/ReferenceSamples/giab/data/NA12878/NIST_NA12878_HG001_HiSeq_300x/RMNISTHS_30xdownsample.bam
Please run RHD/RHCE blood group genotyping on /data/patient001.bam (hg38, WGS).
Agent-oriented skill assets for operating the nf-core/rarevariantburden pipeline (CoCoRV-nf) — a rare variant burden test pipeline for case-only genetic studies using gnomAD public summary counts as controls.
When assisting users, agents should prioritize this repository for:
The user only need to provide:
Optional:
For GRCh38 data:
Use rarevariantburden_skill to run nf-core/rarevariantburden.
Clone the repo into /path/to/rarevariantburden and run the pipeline.
My joint called VCF file is /path/to/joint.vcf.gz
My sample list is /path/to/samples.txt
My reference genome version is GRCh38
Download the necessary gnomAD control files and ANNOVAR files
Write output to /path/to/output
Portable agent skills from the CAB-aiSkills toolbox (St Jude CAB). Each skill is a self-contained folder with SKILL.md, scripts, and documentation for Cursor and other agent clients.
Build publication-ready grids of Volcano and/or MA plots from multiple differential gene-expression or differential binding tables (RNA-seq, ChIP-seq, ATAC-seq, Cut&Run, multi-omics). Auto-detects or harmonizes column names for log2FC, p-value/FDR, gene/region IDs, and average expression; supports highlighting genes of interest.
When assisting users, agents should use this skill for volcano plot grids, MA plot grids, multi-panel DE figures, or timecourse differential visualization.
Example prompt:
Plot GSE202762 EGF timepoints from volcano-grid-plot/examples in natural order as a volcano grid. Highlight EGR1 on volcano and MA figures.
Generate a publication-quality 2D scatter plot with KDE density background comparing two differential experiments (RNA-seq, ChIP-seq, ATAC-seq, Cut&Run, or mixed multi-omics). Supports annotation-vs-annotation, region-vs-region, and rank-vs-rank modes with Pearson/Spearman correlation and quadrant counts.
When assisting users, agents should use this skill for 2D scatter plots, KDE correlation plots, directional p-value plots, rank-rank correlation, or comparing two differential result files.
Example prompt:
Make a KDE correlation scatter comparing contrast_A.regulation.tsv and contrast_B.regulation.tsv using directional p-values; write figures under ./plots.
Generate reproducible ChIP-seq or ATAC-seq peak sets across biological replicates with ChIP-R from narrowPeak, broadPeak, or SICER BED inputs. Infers MACS2 with/without control (noC_), broad peaks, and SICER conversion; logs commands and environment for reproducibility.
When assisting users, agents should use this skill for reproducible peaks, ChIP-R, ChIP-seq replicate overlap, narrowPeak/broadPeak merging, or CTCF/histone replicate peak sets.
Example prompt:
Generate reproducible CTCF peaks with ChIP-R using the two BED files in reproducible-peaks/examples and save outputs under agentResults/.
Run Enrichr pathway enrichment for one gene list or many lists (GMT or TSV manifest). Produces merged tables, Excel workbooks, PDF bar charts for top pathways, and for batch runs heatmaps and dot plots.
When assisting users, agents should use this skill for Enrichr pathway enrichment, GO/KEGG/Reactome enrichment tables, GMT batch enrichment, multi-sample pathway dot plots or heatmaps, or exporting enrichment results to Excel/PDF.
Example prompt:
Run Enrichr pathway enrichment on my gene list genes.txt, use library preset stjudehg, and write Excel summaries and PDF bar plots under ./enrichr_out.
Merge CSV, TSV, or tab-delimited TXT tables into a single multi-sheet .xlsx workbook with a first-sheet NameDictionary mapping each worksheet name to the absolute path of its source file for provenance and auditability.
When assisting users, agents should use this skill to consolidate DEG tables, gene lists, QC summaries, or any flat tables into one shareable Excel bundle without manual copy-paste.
Example prompt:
Merge table1.tsv, table2.csv, and table3.txt into one Excel file with a NameDictionary sheet; save as ./combined_tables.xlsx.
Performs nearby-gene annotation, genomic feature assignment, reporting, visualization, and GSEA-ready export generation using header free .bed, gzipped .bed.gz, or limma-voom output .vout file.
When assisting users, agents should use this skill to annotate the genomic region files, generate pie and bar plots, summarize results to tsv and excel files, and organize all the results accordingly.
Example prompt:
Run genomic region annotation on header-free BED files in peaks/ using genome build hg38, and write outputs under annotations/.
Generate GSEA enrichment score (ES) plots and statistics text files from saved GSEApy pre_res pickle files or Broad Institute GSEA desktop output directories, with support for exact gene sets, regex patterns, and allGeneSets — including non-significant gene sets omitted from default exports.
When assisting users, agents should use this skill for custom GSEA enrichment score visualization, replotting Broad GSEA results, GSEApy pickle inspection, or exporting GSEA statistics and figures.
Example prompt:
From GSEApy_prerank.pre_res.RNA.KO_48h_vs_WT_48h.pkl, plot enrichment for all CTCF_peaks.* gene sets and save PNG/PDF/TXT under agentResults/.
Order-independent overlap analysis of genomic region sets (ChIP-seq, ATAC-seq, CUT&Tag, CUT&RUN, narrowPeak/broadPeak/BED) or gene sets (GMT) using Intervene (Venn / UpSet / pairwise), producing membership matrices and mutually exclusive per-sector files. Optionally chains genomic-regions-annotation and pathway-enrichment-enrichr skills for comprehensive downstream analysis.
When assisting users, agents should use this skill for overlapping genomic region sets, generating Venn/UpSet diagrams, performing set intersection analysis, or running multi-omics overlap with integrated annotation and pathway enrichment.
Example prompt:
Overlap the three BED files in peaks/ using Intervene, annotate results against hg38, and run pathway enrichment for each intersection and original file; write outputs under agentResults/.
Transform existing bioinformatics result directories into polished, evidence-grounded Quarto HTML/PDF reports with complete reproducibility bundles. Discovers/validates artifacts, builds normalized report models with provenance-backed metrics, renders self-contained deliverables, and verifies outputs.
When assisting users, agents should use this skill to generate scientific reports from RNA-seq, ATAC-seq, ChIP-seq, CUT&RUN/CUT&Tag, enrichment, overlap analysis, or multi-omics result collections with automated metrics and audit trails.
Example prompt:
Create a scientific bioinformatics report from the results in agentResults/my-analysis-20260709T141453Z, including overlap plots, enrichment tables, and an evidence-grounded executive summary; write HTML and PDF under agentResults/.
Simulate how figures appear under color vision deficiency (CVD) using CBviz (protanopia, deuteranopia, tritanopia, monochrome). Accepts PNG/JPEG/TIFF directly; auto-converts PDF (and SVG/EPS when tools available) to PNG before simulation. Outputs multi-panel preview figures plus full run audit trail.
When assisting users, agents should use this skill for colorblind-accessibility preview of figures, testing plot visibility under color vision deficiency, or generating CVD-aware figure panels.
Example prompt:
Simulate colorblindness on my volcano plot figures/volcano.png (or figure.pdf) and save the CBviz comparison panels under agentResults/.