Rob Patro

I don't ... remember! but the reads are short (30bp) and all the blast hits I've seen so far have E values > 5!

Ok, this is a first. I have a fastq file with an accession that I *can not find anywhere*. That is, the file has an SRR#, that is embedded in the read headers (and matches the file name), but I can't find this file anywhere (ENA/SRA/etc.). Has anyone ever encountered this?

Current AI agent assessment for coding:

ChatGPT: absolute trash for anything but plotting

Gemini 3 (via antigravity): made nice suggestions, then proceeded to totally break the code

Claude: chatty and hit or miss, but reliably much better than others and often actually useful.

I have to say, Lamar Jackson is an elite quarterback, absolutely! However, this Lamar Jackson (random part of leg/foot), who skips practice all the time and can't throw touchdowns... not a fan of the second dude.

I'll be giving a talk to a CS club at a local high school in January. I welcome any suggestions of topics / activities that people have had success with in such a group. I want to talk about CS (not just AI) and it'd be cool to work in some genomics!

Dear MCPS, I'm closely monitoring your "closely monitoring" nonsense...

Nothing obvious. One thought, perhaps, is that needletail is doing better buffer management than kseq++. It is the case that zlib-rs (the pure rust impl of zlib) is slightly faster than zlib-ng (the fast C impl), but I checked and that's not the difference.

GitHub - bede/deacon: Fast DNA search and [host] depletion using minimizers Fast DNA search and [host] depletion using minimizers - bede/deacon

github.com/bede/deacon
For anyone still using Bowtie2 for filtering or depletion of host sequences or specifics, I can recommend Deacon from @bedec.bsky.social . It is so much faster and easier than Bowtie2, and its performance is equal or better (tested with metagenomes and mitogenomes).🧬 & 🖥️

We now have feature parity in Rust (index construction and reading). Multithreaded decompression and parsing is also about 30% faster than C++! Checkout the mimrs directory in the GitHub repo!

mim: A lightweight auxiliary index to enable fast, parallel, gzipped FASTQ parsing The FASTQ file format is the lingua franca of primary data distribution and processing across most of bioinformatics. Over time, the compression, storage, transmission, and decompression of gzip compr...

This is how we get locked into silly things like fastq.gz. And I realize we’re largely locked in which is why I’m spending some time to try to build backwards compatibility infrastructure that might be able to ameliorate some of the badness: www.biorxiv.org/content/10.1...

I think the format is fundamentally ill suited to purpose in widespread use. As a folk fmt for rapid prototyping of some tools in the 90s, great! But once it seems it will become important, I think it’s silly not to replace it immediately with a proper fmt. This is why we’re stuck where we are. 1/2

a close up of a man 's face in a dark room . ALT: a close up of a man 's face in a dark room .

He’s beginning to believe!

It warms your heart to see one if your favorite devs start to star a bunch of dev-related Rust crates.

We now have a working proof-of-concept implementation of the mim-enabled fastq parser in Rust. It uses needletail for the actual parsing and is *already* ~30% faster than the C++ implementation! Very excited about where this will go! cc @curiouscoding.nl

GFF was originally conceived of as an interchange format for chaining together gene subfeatures (splice sites, promoters, exons, etc) predicted by various specialized tools (HMMs, decision trees, neural nets, SVMs); rather like GENSCAN, but with the idea you could mix & match subfeature predictors

Sometimes, you see something great, and you realize it could be even better with the wisdom of hindsight... and Rust 😅!

"Sure, bedtools is great and happily used by literally everyone, but, it could be bedder" That takes some serious courage. Kudos Brent and @aaronquinlan.bsky.social !

Geriatric millennial here but "weird S thing"? You mean the super awesome, mega stylized, phat as phuck S?

And remember, AI will still produce a ton of slop in 2026, so don't rely on it!

The MCPS school decision today was so bad they needed to apologize. Thanks for the apology, but it's meaningless without real change. You consistently fail MC parents & students with these poor decisions. Please, do better! If you have to subdivide the county into several districts, then do it!

Thanks; they seem to have take a lot from Rust, even the use of (and then move away from) sigils 😋.

What is the carrot syntax?

var reader = BufferedReader(fh^)

What twister group dynamics led to GFF (and GTF), and how can I invent a time machine to prevent these abortions from ever happening?

The Montgomery County, MD school district should be at least 4 separate school districts. Does anyone know the proper people to contact to propose such an initiative?

We are thrilled to announce the first official release (v0.1.8) of #𝗯𝗲𝗱𝗱𝗲𝗿, the successor to one of our flagship tool, #𝗯𝗲𝗱𝘁𝗼𝗼𝗹𝘀! Based on ideas we conceived of long ago (!), this was achieved thanks to the dedication of Brent Pedersen.

1/n

Show me an administration that cares and I'll show you an administration that isn't in power in the US. Wouldn't it be ironic if a federal judge ordered a stop to this, brining about the 10th or so full-blown constitutional crisis of year 1 of the last presidency of the US republic?

After watching that painful Ravens game on Thurs. It's glorious to see the Bills take the Steelers to task!

@profanity.accountant, let's take account.

Don't trust the CDC while RFK Jr. is in charge The HHS secretary, a notorious anti-vaxxer, is now posting his crank views on CDC's website

I can't stay quiet about what RFK jr is doing to the CDC and FDA:
stevensalzberg.substack.com/p/dont-trust...