Issues

• run_pear() ignores config.json's pear.min_overlap, quality_threshold, p_value, and assembly length settings — PEAR always runs with hardcoded defaults

  hassansaei/CapScreen#13 · Filter by author hassansaei opened on Aug 30, 2026

• --cut-umi flag is misleadingly named: it re-trims flanking sequences via Cutadapt but performs no UMI-based PCR-duplicate deduplication

  accuracyenhancement
  hassansaei/CapScreen#12 · Filter by author hassansaei opened on Aug 30, 2026
  

• RPM normalized to mapped reads instead of total input reads makes RPM/enrichment values non-comparable across samples with different mapping efficiency

  hassansaei/CapScreen#11 · Filter by author hassansaei opened on Aug 30, 2026

• Pseudocount (lambda) mismatch between code default (0.01) and shipped config.json (1) compresses enrichment signal for low-abundance variants

  enhancement
  hassansaei/CapScreen#10 · Filter by author hassansaei opened on Aug 30, 2026
  

• Shipped config.json leaves group_roles empty, silently falling back to fragile keyword heuristics that can invert target/background labels

  accuracybug
  hassansaei/CapScreen#9 · Filter by author hassansaei opened on Aug 30, 2026
  

• merge_with_reference() silently discards duplicate ID_WLG entries when peptides collide, misattributing counts

  accuracy
  hassansaei/CapScreen#8 · Filter by author hassansaei opened on Aug 30, 2026

• Zero-tolerance exact peptide matching in merge_with_reference() undercounts variants with sequencing errors

  accuracybug
  hassansaei/CapScreen#7 · Filter by author hassansaei opened on Aug 30, 2026