populationgenomics/talos — reverse-engineered prompt

Build me an open source rare disease genomics reanalysis pipeline called Talos. I want it to take exome or genome variant data for families or cohorts, plus pedigree files and a simple input TSV, then annotate and prioritise variants that could explain a participant’s condition using known disease genes and current evidence.

It should have a preparation step that downloads or formats the big reference resources, ClinVar, PanelApp Australia, AlphaMissense, phenotype data, and anything else needed, then a main run that can handle multisample VCFs, sharded VCFs, or folders of single sample VCFs. The output should be reproducible, scalable, container friendly, and suitable for regular monthly or quarterly reanalysis.

Please include configurable rule based logic aligned with ACMG AMP ideas, inheritance checks, optional phenotype matching, history comparison so newly reportable variants are highlighted, and clear reports for review. Use Python, Nextflow, and Docker, and look up current docs online if you need to.