Skip to content

Support additional minimizer schemes #4

Open

Description

@jianshu93

Hi team,

It worked nicely for my case and I am surprised by the performance for canonical minimizer. I am wondering whether there can be an open and closed syncmer, in the new modmimizer paper, this one is approximating the theoretical lower bound for k < 21 or so.

Best,
Jianshu

Activity

  1. RagnarGrootKoerkamp commented on Mar 25, 2025

    @RagnarGrootKoerkamp
    Contributor

    Oh hey, I see I never replied to this.

    Yeah, adding additional schemes is something that would eventually be nice to do, but it's not very high on my priority list right now.
    Would be happy to supervise a PR on it though.

    To add:

    • mod minimizer, which should be relatively easy: all that's needed is a change of parameter values and a modulo which can be done using fastmod, and since values are small, that should all fit in 32bit operations fine.
    • the ABB+ scheme from here: https://curiouscoding.nl/posts/minimizers/#discussion, which is actually simpler than NTHash, since it just uses a different order on kmers.
    • the open-closed minimizer will be more complicated probably -- haven't really thought about this yet.
  2. changed the title [-]open and closed syncmer with SIMD[/-] [+]Support additional minimizer schemes[/+] on Mar 25, 2025
  3. jianshu93 commented on May 14, 2025

    @jianshu93
    Author

    Thanks for the response anyway. I will close it for now. -Jianshu

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Metadata

Metadata

Assignees

No one assigned

    Labels

    No labels
    No labels

    Type

    No type

    Projects

    No projects

      Milestone

      No milestone

      Relationships

      None yet

      Development

      No branches or pull requests

      Issue actions