fig5-wfmash-query-grid-chop-filter — octopus01:/moosefs/erikg/phrs

Metadata

Status	done
Assigned	`agent-2719`
Agent identity	`46f6237a65ec4f1002c4d3fb201dc8633638d0947c276be7008c227e1051ba5e`
Created	2026-06-24T13:33:08.123344812+00:00
Started	2026-06-24T13:34:48.698074900+00:00
Completed	2026-06-24T13:51:54.662622483+00:00
Tags	`fig5`, `wfmash`, `query-grid`, `sweepga-filter`, `whole-genome`, `eval-scheduled`
Eval score	0.94
└ blocking impact	0.97
└ completeness	0.95
└ constraint fidelity	0.85
└ coordination overhead	0.93
└ correctness	0.96
└ downstream usability	0.95
└ efficiency	0.90
└ intent fidelity	0.82
└ style adherence	0.92

Description

Run the same exact query-grid chop/filter workflow on the updated-bin wfmash -p95 whole-genome PAFs.

Goal: make wfmash comparable to the SweepGA/FastGA query-grid results for both candidate-window panels and whole-genome overview plots.

Inputs:

Raw wfmash PAFs are currently ignored heavy files under /moosefs/erikg/phrs/.wg-worktrees/agent-2636/paper_prep/_brainstorming/pedigree_whole_genome_wfmash_p95_updated_bin/raw_paf/updated_bin_v0.24.2-12-ge040aa10/*.paf.gz
Existing provenance summaries are in paper_prep/_brainstorming/pedigree_whole_genome_wfmash_p95_updated_bin/summaries/.

Required processing:

Verify every raw wfmash PAF has cg:Z on all rows sampled and fail clearly if any row cannot be chopped exactly.
Chop from raw with pafchop-rs --chunk-mode query-grid --overlap 0 for lengths 10000, 5000, 2000.
Filter chopped PAFs with SweepGA PAF filtering: --num-mappings 1:1 --scaffold-jump 0 --scoring ani --overlap 0.
Use distinct output directories under paper_prep/_brainstorming/pedigree_whole_genome_wfmash_p95_updated_bin/query_grid_filter/, e.g. chopped_paf_qgrid_l{N}_o0 and filtered_paf_qgrid_l{N}_o0.
Use Slurm if any step is non-trivial, but these wfmash PAFs are small enough that a short compute-node or carefully bounded local run may be acceptable; do not run an hours-long job on the head node.

Validation/deliverables:

pigz -t all chopped and filtered outputs, write sha256 sidecars.
Write query_grid_filter_manifest.tsv with method=wfmash_p95_updated_bin, comparison_id, chop length, raw/chopped/filtered paths, row counts, commands, binary paths/checksums.
Write candidate-window support summary after filtering, including chr3 retained row count/summed bp/query-union bp for PAN027 and PAN028.
Write README section explaining this is wfmash raw WGA plus the same query-grid/SweepGA 1:1 post-filter used for SweepGA.

Acceptance criteria:

All 9 comparison x length filtered wfmash outputs exist and validate.
Outputs are ready for whole-genome overview plotting.
Commit with message: feat: fig5-wfmash-query-grid-chop-filter (agent-NNN)

Run the same exact query-grid chop/filter workflow on the updated-bin wfmash -p95 whole-genome PAFs.

Goal: make wfmash comparable to the SweepGA/FastGA query-grid results for both candidate-window panels and whole-genome overview plots.

Inputs:
- Raw wfmash PAFs are currently ignored heavy files under /moosefs/erikg/phrs/.wg-worktrees/agent-2636/paper_prep/_brainstorming/pedigree_whole_genome_wfmash_p95_updated_bin/raw_paf/updated_bin_v0.24.2-12-ge040aa10/*.paf.gz
- Existing provenance summaries are in paper_prep/_brainstorming/pedigree_whole_genome_wfmash_p95_updated_bin/summaries/.

Required processing:
- Verify every raw wfmash PAF has cg:Z on all rows sampled and fail clearly if any row cannot be chopped exactly.
- Chop from raw with pafchop-rs --chunk-mode query-grid --overlap 0 for lengths 10000, 5000, 2000.
- Filter chopped PAFs with SweepGA PAF filtering: --num-mappings 1:1 --scaffold-jump 0 --scoring ani --overlap 0.
- Use distinct output directories under paper_prep/_brainstorming/pedigree_whole_genome_wfmash_p95_updated_bin/query_grid_filter/, e.g. chopped_paf_qgrid_l{N}_o0 and filtered_paf_qgrid_l{N}_o0.
- Use Slurm if any step is non-trivial, but these wfmash PAFs are small enough that a short compute-node or carefully bounded local run may be acceptable; do not run an hours-long job on the head node.

Validation/deliverables:
- pigz -t all chopped and filtered outputs, write sha256 sidecars.
- Write query_grid_filter_manifest.tsv with method=wfmash_p95_updated_bin, comparison_id, chop length, raw/chopped/filtered paths, row counts, commands, binary paths/checksums.
- Write candidate-window support summary after filtering, including chr3 retained row count/summed bp/query-union bp for PAN027 and PAN028.
- Write README section explaining this is wfmash raw WGA plus the same query-grid/SweepGA 1:1 post-filter used for SweepGA.

Acceptance criteria:
- All 9 comparison x length filtered wfmash outputs exist and validate.
- Outputs are ready for whole-genome overview plotting.
- Commit with message: feat: fig5-wfmash-query-grid-chop-filter (agent-NNN)

Depends on

Required by

Log

2026-06-24T13:33:57.480942642+00:00 Lightweight assignment: agent=random-agent-46f6237a (46f6237a), exec_mode=full, context_scope=task, reason=Pipeline Operator best matches a bounded command-line data workflow with validation, artifact generation, and manifesting; its concrete-first tradeoff fits the run-and-check nature of this task.
2026-06-24T13:34:48.698078677+00:00 Spawned by coordinator --executor codex --model gpt-5.5
2026-06-24T13:35:00.115204690+00:00 Starting wfmash query-grid chop/filter task
2026-06-24T13:38:38.725350739+00:00 Added reproducible query-grid chop/filter runner and collector scripts; syntax checks starting
2026-06-24T13:39:32.180086819+00:00 Pipeline retry after correcting pafchop binary path to repo target/release
2026-06-24T13:42:44.341128687+00:00 Completed local query-grid chop/filter run and regenerated manifest/candidate summary
2026-06-24T13:48:01.850904811+00:00 Validated: bash -n, python py_compile, cargo test for pafchop-rs, collector revalidation, 9 filtered outputs, 9 chopped outputs, and 18 sha256 sidecars all passed
2026-06-24T13:51:00.929939853+00:00 Committed and pushed: 2a7c510 — feat: fig5-wfmash-query-grid-chop-filter (agent-2719)
2026-06-24T13:51:54.662630358+00:00 Task pending eval (agent reported done; awaiting `.evaluate-*` to score)
2026-06-24T13:59:59.953096892+00:00 PendingEval → Done (evaluator passed; downstream unblocks)