fig5-joint-parent-sweepga

Metadata

Description

Run the corrected direct sweepGA experiment for Fig5 pedigree comparison. The previous direct sweepGA package aligned each child haplotype to parent hap1 and parent hap2 separately, then combined separately filtered PAFs; that is not a joint parent-choice problem and makes 1:1 filtering misleading. Build a new package using one query FASTA for the transmitting child haplotype and one combined target FASTA containing both parental haplotypes for each transmission: PAN027 paternal hap2 vs PAN011 hap1+hap2, PAN027 maternal hap1 vs PAN010 hap1+hap2, and PAN028 maternal hap1 vs PAN027 hap1+hap2. Use the same 500 kb telomeric-window source FASTA (/moosefs/guarracino/HPRCv2/PHR_III/pedigrees/washu/washu.1Mb.telo_500kb_trimmed.fa.gz); do not imply whole-genome alignment and make the 0-500 kb local-window coordinate convention explicit. Submit heavy sweepGA/FastGA alignment through Slurm, not on the head node. Preserve raw many:many/no-scaffold PAFs. Then run joint filtering from the combined raw PAFs for 1:1, 1:many, 2:many, 4:many, and many:many/no-scaffold, all applied jointly across both parent haplotypes. Evaluate whether joint 1:1 still crushes the PHR donor signal, and whether joint 4:many or raw many:many recovers the old untangle-visible chr9q/chr3q and PAR1 structures. Produce exact same-format sibling schematics to compare against paper_prep/_brainstorming/fig5_synteny_recombination_schematic/fig5_synteny_recombination_full.pdf, clearly labeled by filter (at minimum joint 1:1 no-scaffold and joint 4:many no-scaffold). Do not overwrite the old schematic or the direct-sweepGA-1to1 sibling. Include a README explaining: separate-hap filtering was wrong for joint parent choice; local offsets are 500 kb window coordinates; 1:1 is a diagnostic control and may be inappropriate for multimap PHR display; and which filter best preserves the biological signal. Commit and push with WG provenance.

Depends on

• done direct-sweepga-parental
• done fig5-direct-sweepga-1to1-schematic
• done .assign-fig5-joint-parent-sweepga

Required by

• done fig5-whole-genome-joint-parent-sweepga

Log

• 2026-06-20T15:17:00.631568958+00:00 Lightweight assignment: agent=random-agent-46f6237a (46f6237a), exec_mode=full, context_scope=graph, reason=Pipeline Operator is the closest fit for a Slurm-backed alignment/filtering workflow with concrete file updates, raw PAF preservation, and schematic generation across multiple dependent components.
• 2026-06-20T15:17:41.352376132+00:00 Spawned by coordinator --executor codex --model gpt-5.5
• 2026-06-20T15:17:55.970624018+00:00 Starting joint-parent sweepGA package: inspect existing direct concordance outputs, renderer, and task validation requirements.
• 2026-06-20T15:20:53.746329886+00:00 Created joint-parent package skeleton and renderer fork; preparing combined hap1+hap2 target FASTAs next.
• 2026-06-20T15:21:23.170115862+00:00 Prepared joint input FASTAs (46 query windows, 92 combined parent target windows per transmission) and submitted three raw many:many no-scaffold sweepGA jobs through Slurm.
• 2026-06-20T15:23:03.891639341+00:00 Initial Slurm attempt exposed sweepGA multi-target grouping failure; patched target headers to one joint parent genome group and resubmitted jobs 1704274-1704276.
• 2026-06-20T15:23:22.466550774+00:00 Raw joint PAFs validated; ran joint filter matrix from combined raw PAFs and wrote summaries.
• 2026-06-20T15:25:42.905438235+00:00 Validated: Python scripts compile; raw and filtered PAF gzip integrity passes; required joint 1:1 and 4:many PDFs exist; original schematic directories unchanged.
• 2026-06-20T15:28:43.166747612+00:00 Committed and pushed: 5fdd4db — joint-parent sweepGA package, raw/filtered PAFs, summaries, and Fig5 1:1/4:many schematic siblings.
• 2026-06-20T15:29:08.010392975+00:00 Task marked as done