Ragnar {Groot Koerkamp}
@curiouscoding.nl
PhD on high troughput bioinformatics @ ETH Zurich;
IMO, ICPC, Xoogler, Rust, road-cycling, hiking, wild camping, photography
IMO, ICPC, Xoogler, Rust, road-cycling, hiking, wild camping, photography
Reposted by Ragnar {Groot Koerkamp}
Bioinformatics x cybersecurity: Christina Boucher and her colleague Sara Rampazzi uncovered a basic yet critical vulnerability in MinIONs through the MinKNOW software bioengineer.org/portable-gen...
Portable Genetic Sequencer Security Vulnerabilities Could Endanger Personal
Portable genetic sequencers, particularly those manufactured by Oxford Nanopore Technologies, have revolutionized the field of genomics, making DNA sequencing more accessible and practical across the
bioengineer.org
November 12, 2025 at 7:44 AM
Bioinformatics x cybersecurity: Christina Boucher and her colleague Sara Rampazzi uncovered a basic yet critical vulnerability in MinIONs through the MinKNOW software bioengineer.org/portable-gen...
Whenever you make a plot in a paper, please make sure the numbers in there can be interpreted in an absolute sense!
Without being able to extrapolate results, it's impossible to make back-of-the-envelope comparisons to other hardware and tools.
Without being able to extrapolate results, it's impossible to make back-of-the-envelope comparisons to other hardware and tools.
November 12, 2025 at 12:49 AM
Whenever you make a plot in a paper, please make sure the numbers in there can be interpreted in an absolute sense!
Without being able to extrapolate results, it's impossible to make back-of-the-envelope comparisons to other hardware and tools.
Without being able to extrapolate results, it's impossible to make back-of-the-envelope comparisons to other hardware and tools.
Reposted by Ragnar {Groot Koerkamp}
Holy crap! This looks awesome. We've been working toward something just like this (github.com/COMBINE-lab/...), and matchbox seems great; and in Rust! Perhaps we've been scooped, but this looks amazing either way!
November 11, 2025 at 10:04 PM
Holy crap! This looks awesome. We've been working toward something just like this (github.com/COMBINE-lab/...), and matchbox seems great; and in Rust! Perhaps we've been scooped, but this looks amazing either way!
Reposted by Ragnar {Groot Koerkamp}
Powerful read processing with matchbox https://www.biorxiv.org/content/10.1101/2025.11.09.685711v1
November 11, 2025 at 9:47 PM
Powerful read processing with matchbox https://www.biorxiv.org/content/10.1101/2025.11.09.685711v1
Reposted by Ragnar {Groot Koerkamp}
Reposted by Ragnar {Groot Koerkamp}
Come and join us here in Cambridge! Applications open for a new faculty position, for a researcher in computational and/or theoretical biology, based jointly in Genetics and Mathematics. Happy to answer questions about research, teaching and working here.
www.cam.ac.uk/jobs/faculty...
www.cam.ac.uk/jobs/faculty...
Faculty Position in Computational Biology
Applications are invited for an Assistant/ Associate Professorship in Computational Biology to commence on 1 April 2026 or shortly thereafter. This is a joint post between the Department of Applied
www.cam.ac.uk
November 11, 2025 at 9:05 PM
Come and join us here in Cambridge! Applications open for a new faculty position, for a researcher in computational and/or theoretical biology, based jointly in Genetics and Mathematics. Happy to answer questions about research, teaching and working here.
www.cam.ac.uk/jobs/faculty...
www.cam.ac.uk/jobs/faculty...
You know what's really annoying?
The result of summing 2^k 0/1 bits requires k+1 bits to store 😠
Anyway, _mm256_sign_epi32 is my new best friend.
The result of summing 2^k 0/1 bits requires k+1 bits to store 😠
Anyway, _mm256_sign_epi32 is my new best friend.
November 11, 2025 at 12:51 PM
You know what's really annoying?
The result of summing 2^k 0/1 bits requires k+1 bits to store 😠
Anyway, _mm256_sign_epi32 is my new best friend.
The result of summing 2^k 0/1 bits requires k+1 bits to store 😠
Anyway, _mm256_sign_epi32 is my new best friend.
What's the fastest FM-index based mapping tool out there?
How many Gbp/s can it map?
How many Gbp/s can it map?
November 11, 2025 at 10:02 AM
What's the fastest FM-index based mapping tool out there?
How many Gbp/s can it map?
How many Gbp/s can it map?
Reposted by Ragnar {Groot Koerkamp}
We're not planning to have any more children, but if we did, I'd propose to my wife that we name them after an AVX512 instruction, like `VPCLMULQDQ`. That'd give Elmo a run for his money. cc @curiouscoding.nl
November 10, 2025 at 2:06 PM
We're not planning to have any more children, but if we did, I'd propose to my wife that we name them after an AVX512 instruction, like `VPCLMULQDQ`. That'd give Elmo a run for his money. cc @curiouscoding.nl
pdflatex --version: (TeX Live 2026/dev/Arch Linux)
Arch is so bleeding edge, we're already in the future ;)
Arch is so bleeding edge, we're already in the future ;)
November 10, 2025 at 12:26 PM
pdflatex --version: (TeX Live 2026/dev/Arch Linux)
Arch is so bleeding edge, we're already in the future ;)
Arch is so bleeding edge, we're already in the future ;)
If you have a ~GB sized data structure, 2x more or less memory doesn't matter all that much for query latency/throughput.
But if it's ~100kB, then suddenly a factor 2 more or less can be the difference between L1 and L3.
So like, optimizing small metadata is actually more important!?
But if it's ~100kB, then suddenly a factor 2 more or less can be the difference between L1 and L3.
So like, optimizing small metadata is actually more important!?
November 9, 2025 at 11:28 PM
If you have a ~GB sized data structure, 2x more or less memory doesn't matter all that much for query latency/throughput.
But if it's ~100kB, then suddenly a factor 2 more or less can be the difference between L1 and L3.
So like, optimizing small metadata is actually more important!?
But if it's ~100kB, then suddenly a factor 2 more or less can be the difference between L1 and L3.
So like, optimizing small metadata is actually more important!?
damn; computing x^y in floats is at least 100 clock cycles.
Crazy.
Crazy.
November 9, 2025 at 12:15 AM
damn; computing x^y in floats is at least 100 clock cycles.
Crazy.
Crazy.
Reposted by Ragnar {Groot Koerkamp}
I wrote a blog post about Flirt, a new code review tool I started working on:
blog.buenzli.dev/announcing-d...
blog.buenzli.dev/announcing-d...
Announcing Development on Flirt
blog.buenzli.dev
November 5, 2025 at 7:01 PM
I wrote a blog post about Flirt, a new code review tool I started working on:
blog.buenzli.dev/announcing-d...
blog.buenzli.dev/announcing-d...
Reposted by Ragnar {Groot Koerkamp}
I havent read such a banger of an essay in a very long time. This is so well-written!
Maybe Don’t Talk to the New York Times About Zohran Mamdani
It’s remarkable, the people you’ll hear from. Teach for even a little while at an expensive institution—the term they tend to prefer is “elite”—and odds are that eventually someone who was a studen…
lithub.com
November 8, 2025 at 5:37 PM
I havent read such a banger of an essay in a very long time. This is so well-written!
Dutch politics last years: contest who can shout the loudest that everything is an immigration problem.
=> Nobody wants refugee hosting close to them.
> Hotel turns into housing anyway.
> 2 years later: place is closing. People are sad their kids' friends must leave.
www.nrc.nl/nieuws/2025/...
=> Nobody wants refugee hosting close to them.
> Hotel turns into housing anyway.
> 2 years later: place is closing. People are sad their kids' friends must leave.
www.nrc.nl/nieuws/2025/...
Al snel na de komst van de asielzoekers verstomde het protest in Katwoude: ‘We hebben ons gek laten maken’
Asielopvang: In het Noord-Hollandse dorp Katwoude brak twee jaar geleden onrust uit door de komst van een noodopvang voor asielzoekers. Hoe kijken de inwoners er nu tegenaan? „Het zijn mensen zoals ji...
www.nrc.nl
November 8, 2025 at 8:37 PM
Dutch politics last years: contest who can shout the loudest that everything is an immigration problem.
=> Nobody wants refugee hosting close to them.
> Hotel turns into housing anyway.
> 2 years later: place is closing. People are sad their kids' friends must leave.
www.nrc.nl/nieuws/2025/...
=> Nobody wants refugee hosting close to them.
> Hotel turns into housing anyway.
> 2 years later: place is closing. People are sad their kids' friends must leave.
www.nrc.nl/nieuws/2025/...
For memory-bound data structures, the point of fast code is not that the code is fast in itself, but that it uses a minimal number of instructions, so that the out-of-order execution window can start loading memory as far ahead as possible.
November 8, 2025 at 8:31 PM
For memory-bound data structures, the point of fast code is not that the code is fast in itself, but that it uses a minimal number of instructions, so that the out-of-order execution window can start loading memory as far ahead as possible.
Things I want for Christmas: _mm256_srli_epi8
Things I want for Christmas: _mm256_sllv_epi8
November 7, 2025 at 10:46 PM
Things I want for Christmas: _mm256_srli_epi8
computing (1u64<<n)-1 is soo slow and annoying!
LOOKUP_MASK[n] is much better :)
=> another 0.1ns saved in my rank query
LOOKUP_MASK[n] is much better :)
=> another 0.1ns saved in my rank query
November 7, 2025 at 9:39 PM
computing (1u64<<n)-1 is soo slow and annoying!
LOOKUP_MASK[n] is much better :)
=> another 0.1ns saved in my rank query
LOOKUP_MASK[n] is much better :)
=> another 0.1ns saved in my rank query
An average day at the office :)
Currently wishing I could write `Rank` for
Rank usize>, but rust functions cannot be named :(
internals.rust-lang.org/t/pre-rfc-us...
internals.rust-lang.org/t/nameable-f...
Currently wishing I could write `Rank
Rank
internals.rust-lang.org/t/pre-rfc-us...
internals.rust-lang.org/t/nameable-f...
November 7, 2025 at 5:38 PM
An average day at the office :)
Currently wishing I could write `Rank` for
Rank usize>, but rust functions cannot be named :(
internals.rust-lang.org/t/pre-rfc-us...
internals.rust-lang.org/t/nameable-f...
Currently wishing I could write `Rank
Rank
internals.rust-lang.org/t/pre-rfc-us...
internals.rust-lang.org/t/nameable-f...
Every single time I get called, it's my provider trying to sell me bigger subscriptions...
Next time maybe I get an internet-only subscription to get rid of this calling 'feature'.
(If only SMS-based 2FA wasn't a thing anymore :/)
Next time maybe I get an internet-only subscription to get rid of this calling 'feature'.
(If only SMS-based 2FA wasn't a thing anymore :/)
November 7, 2025 at 2:00 PM
Every single time I get called, it's my provider trying to sell me bigger subscriptions...
Next time maybe I get an internet-only subscription to get rid of this calling 'feature'.
(If only SMS-based 2FA wasn't a thing anymore :/)
Next time maybe I get an internet-only subscription to get rid of this calling 'feature'.
(If only SMS-based 2FA wasn't a thing anymore :/)
Wow Rust coroutines are cool!
My little `f(x) { prefetch(x); yield; compute(x); }` benchmark is ~8ns per call with explicit prefetching in a loop, and ~9ns per call when using coroutine abstractions :)
(I first spent a full day struggling with async, which I only got down to 14ns with much pain.)
My little `f(x) { prefetch(x); yield; compute(x); }` benchmark is ~8ns per call with explicit prefetching in a loop, and ~9ns per call when using coroutine abstractions :)
(I first spent a full day struggling with async, which I only got down to 14ns with much pain.)
November 7, 2025 at 9:58 AM
Wow Rust coroutines are cool!
My little `f(x) { prefetch(x); yield; compute(x); }` benchmark is ~8ns per call with explicit prefetching in a loop, and ~9ns per call when using coroutine abstractions :)
(I first spent a full day struggling with async, which I only got down to 14ns with much pain.)
My little `f(x) { prefetch(x); yield; compute(x); }` benchmark is ~8ns per call with explicit prefetching in a loop, and ~9ns per call when using coroutine abstractions :)
(I first spent a full day struggling with async, which I only got down to 14ns with much pain.)
Reposted by Ragnar {Groot Koerkamp}
Our method for genome size estimation from long-read overlaps is now published 🥳
academic.oup.com/bioinformati...
academic.oup.com/bioinformati...
Genome size estimation from long read overlaps
AbstractMotivation. Accurate genome size estimation is an important component of genomic analyses such as assembly and coverage calculation, though existin
academic.oup.com
November 7, 2025 at 3:19 AM
Our method for genome size estimation from long-read overlaps is now published 🥳
academic.oup.com/bioinformati...
academic.oup.com/bioinformati...
I wish GitHub had some feature to send notification mails for repositories/issues I own to me directly, and CC for the rest (everything I'm just watching).
Now there's no distinction at all on what is actionable and what not.
Now there's no distinction at all on what is actionable and what not.
November 6, 2025 at 10:38 PM
I wish GitHub had some feature to send notification mails for repositories/issues I own to me directly, and CC for the rest (everything I'm just watching).
Now there's no distinction at all on what is actionable and what not.
Now there's no distinction at all on what is actionable and what not.
Reposted by Ragnar {Groot Koerkamp}
How on earth do y'all keep track of _which_ federated login system you are using for any given web site??
November 6, 2025 at 3:49 PM
How on earth do y'all keep track of _which_ federated login system you are using for any given web site??
Things I want for Christmas: _mm256_sllv_epi8
November 6, 2025 at 1:38 PM
Things I want for Christmas: _mm256_sllv_epi8