Repository navigation
LNG_CAP_POOLS on by default (deltasignal specs/045); fix CAPPED_IDS cache dependence (F6) - #106
Merged
Merged
Conversation
…d with the decomposition (F6) Past the variant cap, each input set becomes one OR pool instead of a single all-required bundle (RAF's MAP2K/MAPK binding step: 28 AND inputs -> 4). Measured against the same generator commit: held-out 0, experimental +14 (14/0, RAF: ARAF/BRAF/RAF1 now right as KO and as overexpression), curator tuning +21 (21/0); convergence identical. F6: CAPPED_IDS was filled only while decomposing and never cleared, so a cached rebuild emitted no cap pools and ids leaked between pathways. It is now cleared per pathway, written to cache/capped_ids.txt and restored on reuse; a cache without that file is not reused. Runtime tests drive generate_pathway_file (old code fails all 3). Verified: the default build of RAF and Class I MHC equals the measured cap-pools build, and a cached rebuild equals a fresh one. Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
What changes. Past the variant cap (512), each input set becomes one OR pool instead of a single copy that requires every alternative. In RAF's "MAP2Ks and MAPKs bind to the activated RAF complex", that is 28 AND inputs → 4. That step needs 12,000 combinations, and even RAF alone ran out of memory uncapped.
Measured (deltasignal specs/045, pre-registered). Build
20261004-1425_1491276_cappoolsagainstf7ctrl, the same generator commit, scored by the solver with the label-free sweep order:Convergence is identical. ARAF, BRAF and RAF1 are now right both as knockouts and as overexpressions. All of the gain is in RAF, so this is a fix for RAF's all-required bundle, not evidence about the rest of the catalog. No other prediction moved.
Alignment (Adam: the pooled part must line up with its expanded neighbours). The interfaces of the pooled reaction were checked on RAF. A catalog-wide check, which joins copies only where the diagram draws them as one glyph, finds no new break.
F6 fix.
CAPPED_IDSwas filled only while decomposing and never cleared:It is now cleared per pathway, cached in
capped_ids.txtand restored on reuse. Tests intests/test_cap_pools_cache.py; the old code fails all 3. Verified: the default build equals the measured build, and a cached rebuild equals a fresh one (RAF and Class I MHC).🤖 Generated with Claude Code