← graph

fig5-sweepga-1to1-redraw

Fig5 pedigree sweepGA 1:1 inspection redraw

Metadata

Statusdone
Assignedagent-2566
Agent identity46f6237a65ec4f1002c4d3fb201dc8633638d0947c276be7008c227e1051ba5e
Created2026-06-19T14:17:25.407408507+00:00
Started2026-06-19T14:18:44.559527225+00:00
Completed2026-06-19T14:25:59.711414023+00:00
Tagspedigree, fig5, sweepga, redraw, eval-scheduled
Eval score0.94
└ blocking impact0.96
└ completeness0.94
└ constraint fidelity0.10
└ coordination overhead0.96
└ correctness0.95
└ downstream usability0.94
└ efficiency0.89
└ intent fidelity0.91
└ style adherence0.93

Description

Create an author-facing inspection redraw of the pedigree/Fig. 5-style plot using conservative sweepGA-filtered native odgi PAF mappings. This is exploratory output only: do not replace the paper figure and do not edit the manuscript.

Objective: Redo the plot currently used for the pedigree material in the same basic visual/formatting style as the paper's Fig. 5 pedigree untangle panel, but base the plotted intervals on the conservative native-PAF/sweepGA path:

native odgi untangle m1000 n4 PAF -> filter to native nb:i:1 -> sweepGA --num-mappings 1:1 --scaffold-jump 0 -> plot

Context from quick check:

  • Existing native odgi PAFs are in /moosefs/erikg/phrs/pedigree_native_untangle_agent2556_slurm/:
    • PAN027_vs_PAN010.e50000.m1000.j0.8.n4.paf.gz
    • PAN027_vs_PAN011.e50000.m1000.j0.8.n4.paf.gz
    • PAN028_vs_PAN027.e50000.m1000.j0.8.n4.paf.gz
  • Plain sweepGA --num-mappings 1:1 --scaffold-jump 0 technically works, but it still keeps some nb:i:2/3/4 rows because sweepGA can choose equivalent alternates. For the conservative pedigree display, first filter native PAF to nb:i:1, then run sweepGA 1:1 with scaffolding disabled.
  • Preliminary counts from the quick check:
    • PAN027_vs_PAN010: native n4 9810, native nb1 3295, nb1->sweepGA 1:1 no scaffold 2737
    • PAN027_vs_PAN011: native n4 9938, native nb1 3331, nb1->sweepGA 1:1 no scaffold 2835
    • PAN028_vs_PAN027: native n4 8742, native nb1 2948, nb1->sweepGA 1:1 no scaffold 2270
  • Existing quick scratch outputs, if useful, are in /moosefs/erikg/phrs/pedigree_native_untangle_agent2556_sweepga_1to1_noscaffold/ and /moosefs/erikg/phrs/paper_prep/_brainstorming/pedigree_sweepga_1to1_no_scaffold/, but the task should be reproducible and may regenerate them.

Required work:

  1. Inspect the current paper pedigree figure asset and surrounding source/caption to understand its basic formatting and panel logic. Relevant likely files:
    • submission/fig/MainFigures/Fig5_pedigree_untangle.pdf
    • submission/paper.tex
    • any existing pedigree plotting scripts or brainstorming reports under scripts/pedigree/ and paper_prep/_brainstorming/.
  2. Generate sweepGA-filtered PAFs for all three pedigree pairs using the conservative path above. Keep these as external/intermediate files if large; do not commit bulky unnecessary intermediates unless small and clearly useful.
  3. Produce a new Fig. 5-style inspection plot under paper_prep/_brainstorming/fig5_sweepga_1to1_redraw/. Use the same basic visual logic as the paper plot: pedigree pair/panel structure, query/child coordinate orientation, chromosome-arm labels, parental/target colors, and compact layout. It does not need to be publication-final, but it must be readable enough for author review.
  4. Produce at least SVG and PDF if feasible; PNG is useful if easy.
  5. Add a small reproducible script, preferably scripts/pedigree/plot_fig5_sweepga_1to1_redraw.py, that regenerates the plot from the native PAF inputs and records the exact sweepGA command/config.
  6. Add a concise markdown note in the output directory explaining what the plot uses, what was filtered, row counts, and how it differs from plain sweepGA 1:1 and native first-best. State explicitly that this is an inspection redraw and does not update the manuscript figure.

Scope constraints:

  • Do not edit submission/paper.tex, captions, bibliography, or paper figure assets.
  • Do not run heavy odgi graph analysis; use existing native PAFs. sweepGA filtering of the small m1000 PAFs is fine on the head node.
  • Do not introduce scaffolding or repeat-bridging. Use --scaffold-jump 0.
  • Do not let nb:i:2/3/4 rows into the conservative redraw.
  • Preserve ambiguous/multimap findings as context, but make the plotted view the conservative nb1 -> sweepGA 1:1 no scaffold display.

Validation:

  • Script runs from repo root and regenerates the output plot(s).
  • Output directory contains the plot(s), a summary TSV/count table, and a README/note.
  • Summary reports row counts for native n4, native nb1, plain sweepGA n4 1:1 no scaffold, and conservative nb1->sweepGA 1:1 no scaffold for each pair.
  • The conservative plotted PAF contains only nb:i:1 rows.
  • git diff -- submission/paper.tex submission/fig/MainFigures/Fig5_pedigree_untangle.pdf is empty.
  • Commit with WG provenance message, e.g. feat: fig5-sweepga-1to1-redraw (agent-NNN).

Depends on

Required by

Log