Heck, more rewrite-it-in-rust discussions in this ...
# rnaseq_dev
w
Heck, more rewrite-it-in-rust discussions in this channel too - y'all having fun, clearly - any bits need picking up? - I've been pestering #CE9MS66BS for a while now, now it's your turn
j
I'm actually going to encourage a little caution here. As a pipeline maintainer with a community to think about, I have more to consider than the raw speed of any given tool. I need confidence in both the fidelity of a rewrite and its long-term sustainability and maintenance. So I'd gently discourage any lone wolves from jumping in and submitting rewrites unprompted. That said, here's where things stand: - RustQC, by @Phil Ewels with Seqera behind it, is already included as an experimental option. - STAR alignment - the most expensive part of the pipeline — is being rewritten by some of the top experts in the field, so I'm looking forward to that one with confidence. - TrimGalore - as you can see above, @Felix Krueger is already on it. As the original author, he's uniquely placed to match the existing behaviour and support the tool long-term - exactly the kind of continuity I need. For anything remaining, I'm not sure the performance payoff is compelling enough to justify a rewrite by anyone other than the original authors themselves. But the interest is appreciated!
💯 5
You'll have seen Alexander's submission for the duplicate marking above, and I'm not keen on accepting that one right now for the reasons I allude to above. Picard is maintained by one of the foremost genomics organisations in the world and has a lot of community respect. Replacing that is a BIG ask. I suspect that a lot of authors will be taking advantage of AI capabilities to rewrite their own tooling soon, and that's a much better path for pipeline authors.
w
Yeah that's fair enough - especially if any tools are still actively maintained, the original authors should have the first shot at it. (and I probably need to prove myself a bit more on just the one project I've got going)
p
I have a long list of ideas still 😆 #CE8SSJV3N is our most popular pipeline so was an easy first target, but there's a long list after that..
For example, #CFP55NZ5G and #CE5EL6326 both have similar QC subworkflows that run after alignment, I was thinking of extending RustQC with new subcommands to tackle those pipelines (
rustqc chipseq
to complement
rustqc rna
)
Also you'll see an open issue on the RustQC repo about candidates for #CGFUX04HZ