Paper: full-scale R2 head-to-head, warm-cache R3, submitter attribution - #130
Open
davmlaw wants to merge 1 commit into
Open
Paper: full-scale R2 head-to-head, warm-cache R3, submitter attribution#130davmlaw wants to merge 1 commit into
davmlaw wants to merge 1 commit into
Conversation
…on - #112 R2: replace the sampled cdot-vs-UTA comparison with the full-scale head-to-head over all 3,198,528 submitted pairs (cdot 99.0% vs UTA 81.9%; 548,524 resolve through cdot alone, 1,226 through UTA alone), broken down by submission era (2008-2015 cdot 97.3% vs UTA 76.2%). Drops the two-sample framing and the "gated to a sample by UTA throughput" caveat (Table S4). Residual taxonomy recomputed on the full 32,462 residual. R3: replace the "cache conditions differed" hand-wave with a hot-cache measurement: a full-scale local-JSON pass warmed then timed runs at 640 HGVS/s, within 1% of the cold pass (632), so throughput is not sequence-cache-limited (benchmark_fullscale.csv). New: submitter attribution of the versions cdot lacks (submitter_attribution.py) - 94.9% of absent-version citations come from a single large lab and 83.2% of absent versions are cited by one submitter, the self-alignment signature, so substitution not deeper ingest resolves them.
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
🤖 Written by Claude
Follow-up analyses after #128 (which handled items 2, 3 and the review feedback). Three deferred items, all measured on real data:
Item 1 — warm-cache R3 (retires "cache conditions differed")
A full-scale local-JSON pass over all 4,423,358 ClinVar (g.HGVS, c.HGVS) pairs, warmed then timed: 640 HGVS/s hot vs 632 cold (~1%). Full-scale local-JSON throughput is not sequence-cache-limited, so the small local-vs-REST difference was never a cache artifact. R3 rewritten;
benchmark_fullscale.csvadded.Item 4 — full-scale cdot-vs-UTA head-to-head (removes the sampling caveat)
The full submitted corpus (3,198,528 pairs), no sampling:
Broken down by submission era (the fair view, from full data): 2008–2015 cdot 97.3% vs UTA 76.2%; 2021–26 99.0% vs 81.0%. R2 rewritten to full-scale + per-era, dropping the two-sample framing and the Table S4 "gated to a sample by UTA throughput" caveat. Residual taxonomy recomputed on the full 32,462 residual (version-absent dominated).
Item 5 — why is version substitution needed at all? (new result)
Attributing each absent-from-cdot citation to its submitter: of 2,894 labs only 85 ever cite a version cdot lacks, a single large lab accounts for 94.9%, and 83.2% of absent versions are cited by one submitter — the self-alignment signature. So substitution, not deeper ingest, resolves them. Short R2 paragraph +
submitter_attribution.csv(aggregate only; the named per-submitter table is not committed).Facts refreshed:
clinvar_submitted.csv,clinvar_submitted_residual.csv, newbenchmark_fullscale.csv,submitter_attribution.csv. New scriptpaper/scripts/submitter_attribution.py. Paper renders clean.🤖 Generated with Claude Code
https://claude.ai/code/session_015662nbf1iezxdL9EMcckU7