Add a gated parallel domain map that leaves the protein path unchanged - #110
Merged
Conversation
When a query has two or more TED or user domains, run an independent BLAST/Foldseek path into *_domain outputs. Single-domain queries skip that DAG entirely. Co-authored-by: Cursor <cursoragent@cursor.com>
The pandas conda env has no requests, so mocked search-mode CI died in extract_foldseek_hits after it started importing tests.mocks. Co-authored-by: Cursor <cursoragent@cursor.com>
mrubash1
added a commit
to mrubash1/ProteinCartography
that referenced
this pull request
Aug 19, 2026
`make test` is `pytest -vv -s .`, so every invocation drives the whole pipeline three times: search mode, cluster mode, and the domain map added in Arcadia-Science#110. On this machine the default suite does not finish inside ten minutes; with these three deselected it is 61 tests in 0.6 seconds. Two of the three are slow because they do real work. The third, test_pipeline_in_cluster_mode, polls the live public Foldseek server, so its duration is not under this repository's control at all and a contributor with no network cannot run the default suite to completion. This follows the convention the repo already uses for exactly this situation. `live_ted` is registered as a marker and excluded through `addopts`; `slow` is now registered and excluded the same way, so no new command-line flag or conftest hook is introduced. Running them stays a one-liner: pytest -m slow # just these three pytest -m "" # everything, as today Nothing about the tests themselves changes, and CI that wants them can ask for them by name. Co-Authored-By: Claude Opus 5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01Gv8XcYwawycxfLGk4bnz21
mrubash1
added a commit
to mrubash1/ProteinCartography
that referenced
this pull request
Aug 21, 2026
The walkthrough's summary line read "Offline, about two minutes after the conda environments exist, 36 rules." All three claims were wrong, and the first one matters most: a reader who believes the demo is offline will run it somewhere it cannot reach the network and get a failure with no explanation attached. MEASURED, on a cleaned output tree, rather than predicted: wall time 62 s on an 8-core laptop live ted.cathdb.info requests 11, one per input protein dry-run rule count 37 runtime step count 80 `demo/multispace/config.yml` never sets `domain_map` and the default is "auto" (config_utils.py:143), so in cluster mode the domain query gate is handed every .pdb in input_dir -- all eleven. The evidence is left on disk by the run itself: output/domain_path/ted_cache/ holds exactly eleven .ted.json files, ten of which came back with domains, and gate.txt reads "on". THE TWO COUNTS ARE DIFFERENT MEASUREMENTS AND THE README NOW SAYS SO. 37 is rules in the DAG; 80 is steps executed, because several rules run once per protein or per space. The old line said 36, which was neither of them. This is the same confusion CLAUDE.md already flags, now fixed in the document a new contributor actually reads first. This does NOT set `domain_map: "off"`. Turning the demo genuinely offline deletes 43 of its 80 runtime steps, removes the only end-to-end execution of PR Arcadia-Science#110's domain checkpoint anywhere in CI, and moves a DAG count that is pinned in two places. That is a real change with a real trade-off and it is deliberately not being made here; this commit only stops the document from denying what the code does. Documentation only. Unit 1328 passed / 111 skipped / 1 deselected, UNCHANGED -- no test reads this file. ruff check, ruff format (144 unchanged) and snakefmt clean. DAG dry runs 17 / 26 / 37, all unmoved. `git diff --name-only` is this README alone. Co-Authored-By: Claude Opus 5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01SM7RDuLF3HaUcbbf5gchSo
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.
Summary
final_resultsfilenames.min_domain_length, a separate DAG (Snakefile_domain) searches each cropped query domain and writes*_domainmaps. Single-domain queries skip that path (domain_map: auto|off).Test plan
make test(mocked search + cluster pipeline, plus new domain unit/integration tests)test_domain_utils,test_assign_domains,test_aggregate_domain_hits,test_join_domain_features)P00698(1 TED domain): gate off, protein map only, no*_domain.htmlP60709(3 TED domains): protein + domain mapsP02768(5 TED domains): protein + domain mapspytest -m live_tedis opt-in (TED schema check for P00698 chopping22-141).Made with Cursor