ngs-dna-umi-panel-variants

Run the dedicated UMI panel runner with a sample sheet, reference FASTA, target BED, and UMI mode such as duplex. It generates consensus reads, target coverage QC, and variant calls with allele fraction, depth, and strand or duplex support annotations.

Duplex consensus sequencing combines reads from both strands of the same original DNA molecule using paired UMIs, sharply reducing sequencing and PCR errors. This enables confident detection of variants at very low allele fractions in targeted panels.

No, this workflow is designed for panels where molecular barcodes, UMIs, or duplex consensus are central. For ordinary germline calling without molecular consensus, use the ngs-dna-germline-variants skill instead.

The local backend uses fgbio for UMI consensus generation, samtools for alignment processing, and bcftools for variant calling and stats. The preflight script emits an install plan, and missing resources are reported in the readiness outputs.

Trimming UMI bases before their layout and destination are known destroys the molecular barcode information needed for consensus generation. The workflow requires confirming whether UMIs are inline, in index reads, single, or duplex before any read processing.

Key metrics include unique molecular depth versus raw depth, family size distribution, duplex fraction, on-target rate, coverage uniformity, and per-target dropout. The post-run summary and molecular evidence contract track these per sample for low-AF review.