QueryMatcher fix and some small aux data related fixes#1102
Open
mvankem wants to merge 5 commits intosoedinglab:masterfrom
Open
QueryMatcher fix and some small aux data related fixes#1102mvankem wants to merge 5 commits intosoedinglab:masterfrom
mvankem wants to merge 5 commits intosoedinglab:masterfrom
Conversation
…oo full for radix sort
Optionally disable the use of aux seqs in the prefilter.
Author
|
@martin-steinegger this changes the prefiltering, you might want to have a look |
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
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
QueryMatcher: rescore truncated-score diagonals when the buffer is too full for radix sortThis change improves prefilter sensitivity for queries that produce many diagonal hits. When there are more diagonals with truncated ungapped scores (score == 255) than
max-seqs, we need to rescore them to break ties meaningfully. Previously, rescoring + radix sort was only performed when all diagonal hits fit into the double-buffer — both the truncated ones and the ones with scores below 255. When they did not fit, we fell back to a serial sort, and ties between truncated scores were broken arbitrarily without rescoring.The fix first filters out all diagonals with scores below the final
diagonalThr— those are not going to be considered anyway. Rescoring + radix sort is then performed only on the truncated-score subset, which typically fits into the buffer.Add
--aux-scoreflag — Lets the user disable the use of aux seqs in the prefilter.Fix prefiltering query without aux data against targets with aux data — Corrects the aux-data handling when only one side of the search provides aux sequences.