Deduplicate template hits by sequence - #348
Open
etowahadams wants to merge 7 commits into
Open
Conversation
jandom
requested changes
Jul 30, 2026
jandom
left a comment
jandom
left a comment
Collaborator
There was a problem hiding this comment.
Great PR, thank you @etowahadams – there are some minor linting problems
Comment on lines
+2087
to
+2101
| # A2. Deduplicate by ungapped hit sequence, keeping the first | ||
| # (highest-ranked) occurrence. Without this, multiple hits that cover | ||
| # the same query region with identical residues (e.g. several crystal | ||
| # forms of the same complex) each consume a separate template slot. | ||
| if self.deduplicate_sequences: | ||
| if template.seq in seen_sequences: | ||
| if self.create_logs: | ||
| worker_logger.info( | ||
| f"{template.entry_id} {template.chain_id} is a" | ||
| " sequence duplicate of an already-accepted" | ||
| " template." | ||
| ) | ||
| continue | ||
| seen_sequences.add(template.seq) | ||
|
|
Collaborator
There was a problem hiding this comment.
major: let's pull this out into a testable unit of code (a function) and putting under test
| max_release_date: datetime | None = None | ||
| min_release_date_diff: int | None = None | ||
| max_templates: int = 20 | ||
| deduplicate_sequences: bool = True |
Collaborator
There was a problem hiding this comment.
nits: boolean flags are typically a code smell, we should probably wrap this in a configuration object – but this is out of scope for this PR
jandom
reviewed
Aug 5, 2026
Comment on lines
+977
to
+980
| "M8Parser leaves TemplateData.seq unset, so the dedup check in " | ||
| "step A2 compares None while step H records the sequence " | ||
| "recovered by the step E realignment. The check never matches and " | ||
| "deduplicate_sequences is a no-op on the ColabFold path." |
jandom
added
the
safe-to-test
jandom
self-requested a review
August 8, 2026 18:18
jandom
approved these changes
Aug 8, 2026
jandom
left a comment
jandom
left a comment
Collaborator
There was a problem hiding this comment.
After my changes i'm obviously happy with this :D
@etowahadams would be great if you could take another look?
jandom
reviewed
Aug 8, 2026
Comment on lines
+2363
to
+2371
| if self.deduplicate_sequences: | ||
| if template.seq in seen_sequences: | ||
| if self.create_logs: | ||
| worker_logger.info( | ||
| f"{template.entry_id} {template.chain_id} is a" | ||
| " sequence duplicate of an already-accepted template." | ||
| ) | ||
| continue | ||
| seen_sequences.add(template.seq) |
Collaborator
There was a problem hiding this comment.
this is moved significantly more down than it was: depending on the input (a3m, has sequence; m8 no sequence, needs to get seq from cif), the sequence is not available until this step
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
2 participants
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.
During inference, we allow for templates to have the same sequence. This is different from the behavior in AF3, where template hits with the same sequence are filtered out, such that the templates all have different sequences. This PR implements those changes.