review-zoom-v7-hic-mapq0-method-slide

Metadata

Status	done
Assigned	`agent-1126`
Agent identity	`3577bc75d6ed4f1947509aa5c086c91ce7c997c7806dab6bf6affac647452647`
Created	2026-05-07T19:45:16.093526024+00:00
Started	2026-05-07T19:48:34.294123631+00:00
Completed	2026-05-07T20:02:56.288331205+00:00
Tags	`review-zoom`, `review-zoom-v7`, `hic`, `methods`, `mapq0`, `multimapping`, `eval-scheduled`
Eval score	0.96
└ blocking impact	0.97
└ completeness	0.96
└ constraint fidelity	0.85
└ coordination overhead	0.95
└ correctness	0.97
└ downstream usability	0.97
└ efficiency	0.93
└ intent fidelity	0.89
└ style adherence	0.92

Description

Build a concise v7 methods/background slide explaining what we had to do to make the Hi-C / Pore-C / CiFi / Dip-C validation work in subtelomeric repeats. This must run before review-zoom-v7-fanin-render and produce slide-ready material for fan-in.\n\nUser intent:\n- The deck needs to say that the 3D validation required reanalysis of the contact data because standard MAPQ filtering is biased for subtelomeric PHRs.\n- We need MAPQ=0 reads because the signal lives in repetitive subtelomeric DNA; throwing them away discards the data we need.\n- But we do not duplicate multimappers across all possible placements. We handle them carefully: each multimapped read/segment keeps exactly one primary/randomly chosen alignment, adding symmetric noise rather than systematic double-counting.\n\nCanonical report anchors to verify and cite:\n- end-to-end-report/report/05_hic_validation.md, especially Multi-mapping handling lines: Hi-C MIN_MAPQ=0 and RM_MULTI=0; Bowtie2 default no -k/--all; Pore-C/CiFi minimap2 default one primary alignment; no min-mapq filter; consequence is one randomly/arbitrarily chosen placement for equally good subtelomeric mappings.\n- end-to-end-report/report/06_dipc_validation.md: MAPQ=0 throughout sam2seg -q 0 / hickit --min-mapq=0; default MAPQ >=20 would discard 60-99% of subtelomeric reads; BWA-MEM2 reports one primary alignment per read.\n- end-to-end-report/report/07_integrated.md and report/10_limitations.md: flanking unique-sequence control / flanking paradox; signal persists and is stronger in flanking regions, so multimapping is not driving the result, while PHR pair-level contacts remain noisy.\n- subtelomeric_analysis_report.md sections around Hi-C validation, Dip-C, flanking controls, and limitations.\n\nSlide content target:\n- One slide titled something like 'Making Hi-C work at subtelomeric repeats'.\n- A simple 3-step visual/flow:\n 1. Standard unique-map filtering fails here: MAPQ >=20 / removing multimappers deletes most subtelomeric signal.\n 2. Reanalyze with MAPQ=0 and retain multimappers as one primary/randomly chosen placement per read/segment, not all placements.\n 3. Interpret aggregate community signal, not individual repetitive pair contacts; validate with flanking unique-sequence controls and O/E/contact normalization.\n- Include a small plain-language caveat: MAPQ0 adds symmetric noise; it does not support precise pair-level claims in PHR repeats.\n- Include the key phrase that the original standard filtering was biased by ignoring MAPQ0 reads.\n\nDeliverables:\n- README with exact report citations/line anchors and wording constraints.\n- A slide-ready figure or text asset under slides/v2-review-zoom/_revision_assets/v7/hic_mapq0_methods/.\n- SLIDE_PATCH.md telling review-zoom-v7-fanin-render exactly where to insert the slide, likely before the first Hi-C/3D validation slide.\n\nDo not make final deck integration changes; leave that to review-zoom-v7-fanin-render.

Build a concise v7 methods/background slide explaining what we had to do to make the Hi-C / Pore-C / CiFi / Dip-C validation work in subtelomeric repeats. This must run before review-zoom-v7-fanin-render and produce slide-ready material for fan-in.\n\nUser intent:\n- The deck needs to say that the 3D validation required reanalysis of the contact data because standard MAPQ filtering is biased for subtelomeric PHRs.\n- We need MAPQ=0 reads because the signal lives in repetitive subtelomeric DNA; throwing them away discards the data we need.\n- But we do not duplicate multimappers across all possible placements. We handle them carefully: each multimapped read/segment keeps exactly one primary/randomly chosen alignment, adding symmetric noise rather than systematic double-counting.\n\nCanonical report anchors to verify and cite:\n- end-to-end-report/report/05_hic_validation.md, especially Multi-mapping handling lines: Hi-C MIN_MAPQ=0 and RM_MULTI=0; Bowtie2 default no -k/--all; Pore-C/CiFi minimap2 default one primary alignment; no min-mapq filter; consequence is one randomly/arbitrarily chosen placement for equally good subtelomeric mappings.\n- end-to-end-report/report/06_dipc_validation.md: MAPQ=0 throughout sam2seg -q 0 / hickit --min-mapq=0; default MAPQ >=20 would discard 60-99% of subtelomeric reads; BWA-MEM2 reports one primary alignment per read.\n- end-to-end-report/report/07_integrated.md and report/10_limitations.md: flanking unique-sequence control / flanking paradox; signal persists and is stronger in flanking regions, so multimapping is not driving the result, while PHR pair-level contacts remain noisy.\n- subtelomeric_analysis_report.md sections around Hi-C validation, Dip-C, flanking controls, and limitations.\n\nSlide content target:\n- One slide titled something like 'Making Hi-C work at subtelomeric repeats'.\n- A simple 3-step visual/flow:\n  1. Standard unique-map filtering fails here: MAPQ >=20 / removing multimappers deletes most subtelomeric signal.\n  2. Reanalyze with MAPQ=0 and retain multimappers as one primary/randomly chosen placement per read/segment, not all placements.\n  3. Interpret aggregate community signal, not individual repetitive pair contacts; validate with flanking unique-sequence controls and O/E/contact normalization.\n- Include a small plain-language caveat: MAPQ0 adds symmetric noise; it does not support precise pair-level claims in PHR repeats.\n- Include the key phrase that the original standard filtering was biased by ignoring MAPQ0 reads.\n\nDeliverables:\n- README with exact report citations/line anchors and wording constraints.\n- A slide-ready figure or text asset under slides/v2-review-zoom/_revision_assets/v7/hic_mapq0_methods/.\n- SLIDE_PATCH.md telling review-zoom-v7-fanin-render exactly where to insert the slide, likely before the first Hi-C/3D validation slide.\n\nDo not make final deck integration changes; leave that to review-zoom-v7-fanin-render.

Depends on

Required by

Log

2026-05-07T19:45:16.066815166+00:00 Task paused
2026-05-07T19:46:48.661747417+00:00 Task published
2026-05-07T19:48:34.294128250+00:00 Spawned by coordinator --executor codex --model gpt-5.5
2026-05-07T19:48:55.506509893+00:00 Starting documentation asset work; checking task criteria and branch state first.
2026-05-07T19:51:19.194283888+00:00 Verified report anchors for MAPQ0, one-primary multimapper handling, flanking controls, and pair-level caveats; drafting slide assets now.
2026-05-07T20:00:24.953201346+00:00 Validated: README has exact line anchors and wording constraints; slide_text and SVG include MAPQ0, one-placement multimapper handling, aggregate-signal interpretation, and caveat.
2026-05-07T20:00:44.592260162+00:00 Validated: git diff --check passed; SVG compiled through Typst to /tmp/hic_mapq0_svg_test.pdf. Cargo build/test not applicable for docs-only asset task.
2026-05-07T20:02:04.042263994+00:00 Committed: 143dd10 — pushed to remote branch wg/agent-1126/review-zoom-v7-hic-mapq0-method-slide
2026-05-07T20:02:56.288339371+00:00 Task pending eval (agent reported done; awaiting `.evaluate-*` to score)
2026-05-07T20:06:03.742821672+00:00 PendingEval → Done (evaluator passed; downstream unblocks)