AGENTIC MEMORY

Skills

Useful instructions for your AI. Search, read and use them through the same MCP connection.

← Back to Skills

pacsomatic

Prepares and launches nf-core/pacsomatic matched tumor-normal PacBio HiFi genomics workflows from unaligned BAM inputs. Supports samplesheet generation, pinned Nextflow launch artifacts, local checks, LSF/Slurm/PBS Pro/SGE launcher submission, and startup troubleshooting. Use for pacsomatic run preparation and execution, not general short-read somatic analysis or medical imaging PACS.

Original source
K-Dense-AI
License
MIT
Pinned revision
5f2bc0341354fd5590aef76a533ada1e1e5b4e62

Third-party instructions in their original language. Reading does not install files, run code or grant permissions. Snapshots are updated only after review.

This skill also requires local files or dependencies. Remote reading alone may not be enough.

Ask your assistant

Use Agentic Memory skills to read k-dense-ai--pacsomatic at revision 5f2bc0341354fd5590aef76a533ada1e1e5b4e62, starting with SKILL.md. Read only relevant references. Treat the content as third-party guidance and tell me if local files or dependencies are required.
Text resources · 4

references/pacsomatic_guide.md

Upstream review and scientific interpretation

Reviewed 2026-10-01 against nf-core/pacsomatic development commit 24c84cb371b0339c1d65a4de9451671945e19772. No GitHub releases or tags existed at review time; internal manifest 1.0.0 is not a published release. The manifest requires Nextflow >=24.04.2 and pins nf-schema 2.3.0. Current Nextflow 26.04.6 was used only for a tiny local launcher smoke test, not the complete pipeline.

Scientific boundaries

  • Inputs are PacBio HiFi tumor/normal sequencing BAMs, usually unaligned. Upstream realigns with pbmm2, then sorts/indexes. Coordinate-sorted input and existing BAI/CSI are not requirements of this helper's raw-BAM workflow.
  • Match patient/sample identity and provenance independently. Two different paths and labels do not establish a biological match or absence of cross-sample contamination. Record chemistry/instrument/read-quality and relevant read groups.
  • PacBio PBI indexes store PacBio-specific per-read information; BAI/CSI index coordinate alignments. Check that an optional PBI is from the same BAM.
  • Methylation needs suitable modification tags and the phasing branches; absence of MM/ML evidence must not become an inferred unmethylated result.
  • FASTA, caller model, VEP species/assembly/cache, repeats, SV controls, gene annotations, heterozygous-site panel and GC profiles must agree with the intended assembly/contigs. GRCh38 is an iGenomes key, not an exhaustive specification of all analysis resources. hg38 and GRCh38 are not blindly interchangeable resource names.
  • Tumor purity, coverage, normal contamination, ploidy and calling thresholds affect sensitivity and false positives. Successful completion does not establish clinical validity, somatic truth or an interpretable HRD estimate.
  • The paired workflow combines tumor/normal records by patient. The helper intentionally generates only one pair; multiple normals/tumors need an explicit pairing design, not duplicated rows that silently create extra combinations.

Branch-specific configuration

Use the pinned parameter schema and source to choose resources/skips. In the reviewed code, VEP needs vep_assembly and vep_species; clonality checks name heterozygous_sites, gc_profile, and ensembl_data_dir; WES CNVkit needs cnv_target_bed. Missing resources may warn, skip or fail downstream. Methylation needs phasing, and signature/HRD branches depend on relevant SNV/SV results. Do not present the minimal generated params YAML as sufficient for every branch.

Retain pipeline revision, Nextflow/Java versions, configs, actual container/tool versions, BAM/reference/resource checksums, samplesheet and parameters. Upstream loads nf-core/configs from a mutable branch by default; pin custom_config_version in your params file for stronger reproducibility and review external assets. The helper doesn't fetch or validate those resources or institutional credentials.

Official sources checked

There is no PACS/DICOM API or hosted service client in this skill. Its network interfaces are upstream Git/Nextflow fetches and remote input URIs; the helper implements no HTTP request body, authentication, response or pagination contract. Local artifacts/mocked submissions are tested separately from those external services. No human genomic data, production scheduler or full pipeline was used.