Skip to content

Report: record BV-BRC-only evidence as not available, not zero, on non-genome runs #90

Open

Description

@Jiahuic

Goal: a sequence-only report must never look like a full genome report. Part of #87.

When BV-BRC is not the source, many fields are not "zero" but "not available": specialty genes, essentiality, taxonomy lineage, genome metadata. A score component with no evidence behind it currently contributes 0, exactly like "checked and false" (pitfall #12).

Scope

  • Every BV-BRC-derived field in report.json records its source, or not_available with a reason.
  • Extend the runner warning added in Runner: --human-homology and --essentiality cannot be reached from the pipeline #81 to cover all BV-BRC-only components, not just human homology and essentiality.
  • The HTML report shows a per-protein "evidence coverage" line: which sources answered, which did not apply, which were not run.

Definition of done

  • A --from-fasta run lists every unavailable source in run.json and on the console
  • The report renders "not available" distinctly from "none found"
  • A test pins that a genome run and a FASTA run of the same protein are distinguishable from report.json alone

Unassigned on purpose: comment to claim this task, then assign yourself.

Activity

  1. Jiahuic commented on Oct 9, 2026

    @Jiahuic
    CollaboratorAuthor

    Status: #101 implements the run-level part (open, ready to merge, after #98 → #99 → #100).

    Done:

    • On the --from-fasta route, M2 records not_available (component → reason) in m2_pdb/run.json. The runner prints those components as "not available", and the full report shows a "Not available on this run" note in its header.
    • A genome run and a FASTA run of the same protein are distinguishable from report.json alone (genome.annotation_route, genome.not_available), and a test pins this.

    Left before closing:

    • A per-protein "evidence coverage" line in the HTML report: which sources answered, which did not apply, which were not run. The run-level note covers the FASTA route today, because every protein there lacks the same sources. A per-protein line matters once sources differ by protein: a partial UniProt mapping, a provider that failed for some proteins, or M2: virulence and AMR evidence without BV-BRC — VFDB and CARD by DIAMOND #91's VFDB/CARD rows.
    • Extend not_available beyond the FASTA route if other routes turn out to lack sources. None are known yet.
  2. added a commit that references this issue on Nov 15, 2026
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Metadata

Metadata

Assignees

No one assigned

    Type

    No type

    Projects

    No projects

      Milestone

      No milestone

      Relationships

      None yet

      Development

      No branches or pull requests

      Issue actions