Jim Shaw
@jimshaw.bsky.social
Postdoc at Dana-Farber and Harvard Med with Heng Li (@lh3lh3.bsky.social). Prev: UBC / UofT.
I like thinking about biological sequence analysis and its applications to metagenomics / microbial genomics.
https://jim-shaw-bluenote.github.io
I like thinking about biological sequence analysis and its applications to metagenomics / microbial genomics.
https://jim-shaw-bluenote.github.io
Great to see that Emu's still being worked on and updated!
Indeed, I think Emu and savont will serve different niches in the future and its very cool to see your development. Thanks for the informative post.
Indeed, I think Emu and savont will serve different niches in the future and its very cool to see your development. Thanks for the informative post.
February 13, 2026 at 9:07 PM
Great to see that Emu's still being worked on and updated!
Indeed, I think Emu and savont will serve different niches in the future and its very cool to see your development. Thanks for the informative post.
Indeed, I think Emu and savont will serve different niches in the future and its very cool to see your development. Thanks for the informative post.
Reposted by Jim Shaw
I enjoyed reading your post @jimshaw.bsky.social and learning about Savont, thank you for the comparison to Emu and kudos on your new tool!
relevant to points raised in the post, the Emu development team has compiled our follow-up thoughts here: github.com/treangenlab/...
relevant to points raised in the post, the Emu development team has compiled our follow-up thoughts here: github.com/treangenlab/...
Recent updates
Contribute to treangenlab/emu development by creating an account on GitHub.
github.com
February 13, 2026 at 8:02 PM
I enjoyed reading your post @jimshaw.bsky.social and learning about Savont, thank you for the comparison to Emu and kudos on your new tool!
relevant to points raised in the post, the Emu development team has compiled our follow-up thoughts here: github.com/treangenlab/...
relevant to points raised in the post, the Emu development team has compiled our follow-up thoughts here: github.com/treangenlab/...
Reposted by Jim Shaw
Now also on arxiv:
arxiv.org/abs/2602.04103
arxiv.org/abs/2602.04103
QuadRank: Engineering a High Throughput Rank
Given a text, a query $\mathsf{rank}(q, c)$ counts the number of occurrences of character $c$ among the first $q$ characters of the text. Space-efficient methods to answer these rank queries form an i...
arxiv.org
February 5, 2026 at 10:53 AM
Now also on arxiv:
arxiv.org/abs/2602.04103
arxiv.org/abs/2602.04103
Awesome to hear, let me know how it goes.
January 28, 2026 at 11:39 PM
Awesome to hear, let me know how it goes.
Work supervised by @lh3lh3.bsky.social and in collab with @mkddueholm.bsky.social and his wonderful team.
Savont is still quite new, so feedback is more than welcome.
4/4
Savont is still quite new, so feedback is more than welcome.
4/4
January 28, 2026 at 6:47 PM
Work supervised by @lh3lh3.bsky.social and in collab with @mkddueholm.bsky.social and his wonderful team.
Savont is still quite new, so feedback is more than welcome.
4/4
Savont is still quite new, so feedback is more than welcome.
4/4
Using ASVs seems to avoid some issues with low-abundance, false-positive species calls from read mapping. Of course, ASVs will be less sensitive, though.
For more preliminary results and tests on actual data, see github.com/bluenote-157...
3/4
For more preliminary results and tests on actual data, see github.com/bluenote-157...
3/4
January 28, 2026 at 6:46 PM
Using ASVs seems to avoid some issues with low-abundance, false-positive species calls from read mapping. Of course, ASVs will be less sensitive, though.
For more preliminary results and tests on actual data, see github.com/bluenote-157...
3/4
For more preliminary results and tests on actual data, see github.com/bluenote-157...
3/4
Preliminary: savont gets ASVs with ~10-30x less depth than previous denoising methods for *nanopore R10.4 amplicons*.
Essentially, previous short-read methods have difficulties with longer reads and higher error rates. Savont uses new techniques to handle this.
2/4
Essentially, previous short-read methods have difficulties with longer reads and higher error rates. Savont uses new techniques to handle this.
2/4
January 28, 2026 at 6:46 PM
Preliminary: savont gets ASVs with ~10-30x less depth than previous denoising methods for *nanopore R10.4 amplicons*.
Essentially, previous short-read methods have difficulties with longer reads and higher error rates. Savont uses new techniques to handle this.
2/4
Essentially, previous short-read methods have difficulties with longer reads and higher error rates. Savont uses new techniques to handle this.
2/4