1/ Our paper on Multi-Context Seeds is now out, with @tolyan.bsky.social spearheading the work and contributions from Nicolas and @marcelm.net. We introduce a new seeding concept that improves read alignment accuracy while maintaining speed.
link.springer.com/article/10.1...
Als politici zeggen dat ze nog niet weten of de VS en Israël het internationaal recht hebben geschonden, ben ik wel benieuwd waarom ze 2 maanden terug, 5 dagen na de operatie tegen Maduro, zeker wisten dat dit recht toen wel was geschonden. Ook D66, VVD en CDA
www.tweedekamer.nl/downloads/do...
Since it's not on @biorxivpreprint.bsky.social, my normal crowd might not see it, but I'm very excited about this recent work that, actually, started as class project in my graduate genomics class last year: eprint.iacr.org/2026/463 . Zero knowledge proofs for genomics; with algorithms :).
1350 train kms in 16h for 40€ ✅
Excited to share this preprint that describes my latest work on using GPUs to accelerate processing of RNA-seq data.
The title says it all: "RNA-seq analysis in seconds using GPUs" now on biorxiv www.biorxiv.org/content/10.6... and github github.com/pachterlab/k...
Figure 1 shows they key result
Israel is bombarding, literally bombarding, two Middle East capitals, Beirut and Tehran, killing 100s of civilians, and yet the US and UK media continue to portray Iran as the threat to the region.
Israel has nukes, but Iran is the nuclear threat.
We live in Orwellian times.
Yeah, ignorance is bliss 🤩
Just in academia that stops working once you want to make a paper
> The compromised version was live for eight hours before StepSecurity's automated monitoring flagged it
did they publish /exactly/ at the time of the only oncall person's bed time and did they have a super restful night?
(3/3)
What?!? How can one be 200x slower than the other....
Btw, saw it also mentioned on reddit, but what about plotlars?
Tthese separators follow landmarks. Eg the set of bridges over the Rhine is an efficient way to partition Europe into two halves. Or passes over the alps, and so on.
This partitioning does not depend on edge weights, so that it's relatively cheap to recompute paths in case of eg traffic jams.
Basically, we take a set of nodes as the root 'super node' that we turn into a clique, such that the remainder is partitioned into two (or more) connected components. Then we can recurse to get a tree of 'separating sets'.
In practice, these separators are small, giving O(n^{1/3}) queries.
The tl;dr of CCH: in a binary tree, you can find the distance between any 2 nodes s and t by computing d(s, u) and d(v, t) for all ancestors u of s and ancestors v of t. Then find the LCA and sum the two. So lg(n) query time.
Road graphs are not trees, but we can 'contract' them into one.
The EU road network used for testing has 18M nodes and 42M edges. Dijkstra between two random nodes on this graph takes around a second.
The existing C++ implementations of CCH take only ~100us, which is already insane.
With some SIMD tricks, mine brings it down to ~40us.
I got nerd-sniped into writing some code for Customizable Contraction Hierarchies, which is an insanely cool and super simple data structure for fast route planning.
Nice survey paper by Michael Zündorf et al: arxiv.org/abs/2502.10519
My unpolished implementation nodes: curiouscoding.nl/posts/cch/
If you ever need to get a refund on NordVPN auto-renewal, you just need to add ?intent_rf=true to the URL of the billing page and suddenly a 'refund' button will appear...
(Took way too long of browsing reddit to figure that out.)
curiouscoding.nl/posts/nordvp...
Oh cool 🥰
start-history.com graph of A*PA github stars, with bumps at time of preprint and publication, and steady growth since then.
Are people actually using A*PA?
It suddenly got 4 stars in the last 2 days and appears to be growing steadily either way, but I'm getting issues maybe twice a year, which can either mean nobody uses it, or that it just works (but the UX is really not so great).
Broke a leg playing table tennis. Got some titanium goodies now :)
[T]::array_windows is stable!!!
I'm afraid I don't know most of those words; will check it out :)
Yey for sans serif!
Time for the rise of super small and efficient STPD-index based aligners!
What?! do they think it's 2016 still??
Can barely have a human genome in memory?!
Mostly specializing all the transpose, collect and dedup functions for it. Idk how easy that is (/how similar it is to other arches)
The main simd-mini issue was basically that building for 32bit targets means that usize is not 64 bits and this was implicitly assumed in a few places.
And there was a silly build error for all non-simd target unrelated to that. Although I guess proper simd support for wasm would be cool.
When someone says „Scientists do not want you to know“ you can dismiss everything from there on. Scientists want you to know. They are desperate that you know. They can’t shut up about what they found out and want you to know.
@fingels.bsky.social sounds familiar
Mer de Glace from Signal Forbes
1890 | 2013 | 2021 | 2024
Strong acceleration of the demise of the largest French glacier over last few years! 🧊🔥
Deacon can now run in the browser using WebAssembly. Sequence data never leaves your machine. It currently supports FASTA/Q filtering using indexes up to 1GB in size.
Demo: bede.im/deacon
soms weet ik zelf ook niet zeker of ik een tv-recensie of een speld-artikel aan het schrijven ben
