bio-skills

Plugin bio-skills from marketplace sahuno in sahuno/llm_configs. Commands read from the repo's manifest — 12 days ago.

Skills (13)

• analysis-gotchas

  Empirically confirmed failure modes in genomics analysis tools that fail SILENTLY — producing plausible-looking output that is wrong. Use this skill proactively whenever the work involves: DSS / DMLtest / differential methylation, parallel R (mclapply, cv.glmnet, BSseq, Seurat per-sample loops) on S

• analysis-playbooks

  The lab's standard analysis chains, QC checkpoints and conventions, kept out of CLAUDE.md so they load only when the work calls for them. Use when planning, running or reviewing: ONT methylation (pod5, dorado basecall or align, modkit pileup / dmr, DMRs); variant calling (Clair3, Sniffles2, GATK Hap

• chimeric-read-validation

  Check whether structural-variant or breakpoint calls (Severus, Manta, Sniffles2, Delly, GRIDSS, MELT, Arriba, SvABA) are backed by real chimeric reads: viral integrations, gene fusions, mobile-element insertions, translocations. Use when asked to validate, QC or trust SV/fusion/integration calls, ab

• cohort-overview

  Build a per-patient × per-assay sample-availability heatmap (SPECTRUM-style cohort overview) from a wide TSV. Patient-level annotations (signature, BRCA1/2, CCNE1) on top, site strip on left, three-state cells (Yes/No/NA) with customizable colours and optional NA→No collapse. Use proactively when th

• docker-hpc

  Write Dockerfiles for scientific and bioinformatics tools (R/Bioconductor, conda/pip, CUDA) that Apptainer will pull on HPC, plus GitHub Actions that build and push to Docker Hub with no local Docker daemon, in a monorepo of containers. Use for 'dockerize X', 'build a Docker image for Y', 'container

• igv-screenshots

  This skill generates IGV screenshots using igver for genomic visualization. Use when user asks to "take IGV screenshot", "IGV snapshot", "visualize BAM in IGV", "igver screenshot", "show me the reads at", "screenshot genomic region", "visualize methylation in IGV", "IGV plot of BAM", "generate IGV i

• journal-club

  Phased workflow for analysing and presenting a research paper: ingest, comprehension quiz, causal claims with quoted evidence (necessity vs sufficiency vs not proven), statistics and reproducibility audit (sample sizes, multiple testing, data deposition, RRIDs, code), critique, slide outline and dra

• nfcore-module

  End-to-end builder for new nf-core modules. Scaffolds all required files, runs lint and nf-test in a loop until both pass, and produces PR-ready artifacts (description, Slack draft, checklist). Use this skill proactively whenever the user wants to: create a new nf-core module, add a tool to nf-core/

• runtime-resource-study

  Stage-gated runtime and memory study for a command-line tool on SLURM: preflight, one-factor scans, a 2^k factorial, implementation comparison, input-size scan and validation, ending in a fitted resource model (wall time and peak RSS vs input size), model.yaml, REPORT.md and a one-page PDF. Use when

• scatter-gather

  Decide whether a genomics job can safely be split across chromosomes or region tiles, how to shard it, and how to merge per-shard BAM/VCF/BED/bedMethyl/bigWig outputs without silent corruption (some methods, e.g. DSS DMLtest, pool genome-wide and break when sharded). Use for parallelising per chromo

• simplified-technical-english

  Rewrite or score text in ASD-STE100 Simplified Technical English, with a strictness dial (strict, standard ~80%, light, custom). Use when asked for STE, ASD-STE100 or controlled English, or to simplify a procedure, SOP, protocol, manual, safety notice or README for non-native English readers. Prefer

• singularity-build

  Build Singularity/Apptainer containers (.def files, .sif images) for bioinformatics tools on MSKCC HPC without sudo: --fakeroot, conda not apt-get, SLURM-safe build scripts; first tries to pull and verify a pre-built image. Use when asked to containerize a tool for Apptainer, write a .def file, fix

• snakemake

  Write, debug and run Snakemake 9 workflows on SLURM: Snakefiles and rules, SLURM executor and workflow profiles, containers, run-directory layout, and failures (memory conflicts, container propagation, stale locks, failed jobs). Use when asked to build or fix a Snakemake pipeline, add rules, run a d

Slash commands (10)

• /bio-skills:add-gotcha

  Record a tool failure mode from this session as a durable gotcha record

• /bio-skills:figure-manifest

  Record or check which script, commit and inputs produced each figure in a run. Use when the user asks which figures are final or ready to report, which script made a figure, or wants figures traced to their code before a paper, talk or revision.

• /bio-skills:gates

  Run the pre-completion quality gates before declaring any analysis done

• /bio-skills:gotcha-audit

  Check the gotcha records for staleness, missing versions, and missing detection commands

• /bio-skills:init-bio-project

  Scaffold a genomics project directory (config, sample sheet, immutable raw/, genome-tagged processed/, workflow dirs)

• /bio-skills:preflight

  Before scaffolding any pipeline, check whether a validated implementation already exists

• /bio-skills:promote-raw

  Promote staged data from data/inbox/ to the immutable data/raw/, with checksums and provenance

• /bio-skills:review

  Adversarial review of code written this session — data integrity, genome safety, pipeline correctness, statistical rigor, reproducibility

• /bio-skills:verify-run

  Verify a completed run actually succeeded — exit 0 is not evidence

• /bio-skills:wrapup

  Append this session's work to the project ledger (PROGRESS.md) so the next session can resume without re-discovery

Subagents (3)

• repro-auditor

  Audits whether an analysis could be rerun from the repository alone by someone who was not there. Invoke before archiving a project, before submitting a manuscript, when handing work to a collaborator, or when the user asks whether work is reproducible.

• reviewer-2

  Adversarial reviewer for a scientific claim, figure, or draft. Invoke before submission, before presenting to collaborators, or whenever a result feels ready — its job is to attack the claim the way a hostile referee would. Use proactively when the user says a finding is "done", "ready", "significan

• stats-reviewer

  Reviews the statistical choices in an analysis — test selection, multiple testing, effect sizes, sample size, cross-validation at small n. Invoke when a result rests on a p-value, when choosing a test, before reporting any significance claim, or when the user asks whether an analysis is statisticall

Similar plugins