← graph

fig5-whole-genome-wfmash-p95-evidence-review

Review wfmash -p95 Fig5 raw PAF evidence

Metadata

Statusdone
Assignedagent-2633
Agent identity289ccc9f03fc7c121a5ab8d685ffd018371bcdac67ceab1d50b03e7347d29155
Created2026-06-21T06:59:07.750329707+00:00
Started2026-06-21T08:45:24.901099974+00:00
Completed2026-06-21T08:51:07.555513004+00:00
Tagspedigree, fig5, wfmash, evidence-review, chr3-homology, whole-genome-alignment, raw-paf, eval-scheduled
Eval score0.95
└ blocking impact0.97
└ completeness0.96
└ constraint fidelity0.85
└ coordination overhead0.94
└ correctness0.96
└ downstream usability0.96
└ efficiency0.92
└ intent fidelity0.88
└ style adherence0.96

Description

Input context:

  • New wfmash package from fig5-whole-genome-wfmash-p95: paper_prep/_brainstorming/pedigree_whole_genome_wfmash_p95/.
  • Prior sweepGA evidence review: paper_prep/_brainstorming/fig5_whole_genome_sweepga_evidence_review/REPORT.md.
  • Prior QA closeout: paper_prep/_brainstorming/fig5_whole_genome_sweepga_closeout/QA_REPORT.md.

Task: Review the whole-genome wfmash -p 95 raw PAF results for Fig5. The key question is whether wfmash recovers chr3-target homology overlapping the PAN027/PAN028 chr9 candidate windows that full-genome sweepGA/FastGA did not emit, before any filtering.

Review requirements:

  • Audit that the wfmash execution used full whole-genome inputs for the same three comparisons and did not replace them with window-only/chromosome-only alignments.
  • Audit exact wfmash commands and confirm -p 95 was used.
  • Check whether /dev/shm or node-local scratch was used through wfmash -B or an equivalent documented scratch path.
  • Inspect raw PAF/chopped only posthoc: do not treat a filtered or one-to-one subset as the primary result.
  • Recompute or spot-check summaries/candidate_window_support.tsv from raw PAFs. Specifically state whether chr3-target rows overlap the PAN027 and/or PAN028 chr9 candidate windows and report target contig names, query/target intervals, strand, alignment length, identity/score fields available, and overlap with the candidate window.
  • Compare against prior sweepGA raw many:many -j 0 absence of chr3 rows. Make clear whether wfmash changes the evidence status for Fig5.

Deliverable: Create paper_prep/_brainstorming/fig5_whole_genome_wfmash_p95_evidence_review/REPORT.md plus any small TSV summaries needed for the report.

Acceptance:

  • The report gives a direct yes/no answer: did raw whole-genome wfmash -p 95 recover chr3 support for the Fig5 chr9 candidate windows?
  • The report separates technical provenance from biological interpretation.
  • If wfmash recovers chr3 support, the report identifies the minimal raw PAF rows that should be used for any future schematic; if it does not, it states that no direct whole-genome wfmash support was found.
  • No submission/ files are changed and no Fig5 schematic is created in this review task.

Depends on

Required by

✉ Messages 1 message (1 unread)

  1. #1codex2026-06-21T07:08:10.855113898+00:00delivered
    Version/provenance requirement added after user check: local Guix wfmash is stale (`0.12.5-1+0222f7c`), while upstream official GitHub latest checked from chat is `v0.24.2`. In the evidence review, require that the primary wfmash PAFs come from a current upstream release/build (or explicitly explain why not). If outputs from old 0.12.5 exist, treat them as legacy diagnostics only, not final Fig5 evidence. Also audit current CLI semantics; do not assume old `-s` segment-length meaning on v0.24.x.

Log