@XLR

Husband. Parent. Dr sc @ETH. Bioinformagician. Father of @PacBio HiFi. Fellow BFX/AI. C++. Geek. Digital Artist. Electronic music. Speaks in memes.

 Germany
armintoepfer.com
Joined November 2008
  • Tweets7.3K
  • Following496
  • Followers1.5K
1.9KMedia
After 12 years at @PacBio, I'm proud to announce that I've been promoted to Fellow - moving back to a full-time individual contributor role to work with the great folks across the company! I'm excited to finally have the space to explore the hard problems and even crazier ideas
11
893
With a minimal CLAUDE. md and guards to validate outputs, I can talk to Opus like to a team of the best SWEs in my field. It will outperform any of us. Still needs plenty of steering that I'm willing to provide.
1
4
362
Max Q! I slowed down the pace to focus on memory usage: - 40 seconds for unaligned CRAM (-10s) - 7 GB RAM for unaligned BAM (-11GB) Comparing against pbmm2 and the last checkpoint. Pushing the frontier of bioinformatics is addictive. 2026Q4 shapes up to be banger!
Armin Töpfer@XLROct 8
Lift-off! Aligning 30x HiFi HG002 in less than 60 seconds on 64 cores. - 128c: 33s/59s un/sorted CRAM - CRAM output reduces output size by 2.5x - Better SNP, INDEL, SV F1s - Lower peak memory It's a PoC for now! @lh3lh3 I'll start porting some of it to mm2 soon.
1
5
428
Developing new software in an established field requires constant output testing. Those gates must be FAST to not slow down the development cycle. DeepVariant 40s, hap.py 3s, Truvari 3.5 minutes? No longer! How about 19 seconds with identical results.
3
4
16
1,173
Me every morning to check what my army of agents produced over night.
1
148
Trying to keep up with open-weight models...
116
Lift-off! Aligning 30x HiFi HG002 in less than 60 seconds on 64 cores. - 128c: 33s/59s un/sorted CRAM - CRAM output reduces output size by 2.5x - Better SNP, INDEL, SV F1s - Lower peak memory It's a PoC for now! @lh3lh3 I'll start porting some of it to mm2 soon.
Armin Töpfer@XLROct 7
How much faster can we map and align accurate long reads? I benchmarked pbmm2 on 30× HiFi HG002 against GRCh38 using 64 CPU cores: ~17 minutes. Can we get below 2 minutes while maintaining SNV, INDEL, and SV calling metrics? mm2 was a jump. Is it time to leap?
3
8
2
39
2,998
And another experiment, re-implemented hap.py in C++ to get it from 16 min to 3 seconds. Identical results. What's next?!
1
1
3
195
How much faster can we map and align accurate long reads? I benchmarked pbmm2 on 30× HiFi HG002 against GRCh38 using 64 CPU cores: ~17 minutes. Can we get below 2 minutes while maintaining SNV, INDEL, and SV calling metrics? mm2 was a jump. Is it time to leap?
2
1
1
10
3,073
How far we've come in bioinformatics! Google won Truth Challenge in May 2016. Then Parabricks (later Nvidia) made it significantly faster. Now it's a commodity. Cheers to everyone maintaining it for the past decade! Now that it's so damn fast, can we make it better?
Armin Töpfer@XLROct 5
This weekend was fun, as I now have a PacBio DeepVariant caller with < 1 minute TAT 🎉 Even pure CPU is faster than what others are offering on GPU.
2
13
727
This weekend was fun, as I now have a PacBio DeepVariant caller with < 1 minute TAT 🎉 Even pure CPU is faster than what others are offering on GPU.
Armin Töpfer@XLROct 4
This weekend's Opus /goal was simple and crazy. Run DeepVariant variant calling on a 128c + 1 RTX 6000 Blackwell server in less than 60s.
4
2
1
17
1,659
Next goal: Finish a trio – both parents and child at 20x HiFi – in less than one minute. Let's make variant calling dirt cheap and fast.
1
130
Done after talking to Claude on my phone for 23 hours. 30-fold HiFi called all variants in 52 seconds, aligned BAM to VCF. How insane is that!
Armin Töpfer@XLROct 4
This weekend's Opus /goal was simple and crazy. Run DeepVariant variant calling on a 128c + 1 RTX 6000 Blackwell server in less than 60s.
1
3
491
This weekend's Opus /goal was simple and crazy. Run DeepVariant variant calling on a 128c + 1 RTX 6000 Blackwell server in less than 60s.
1
1
1
1,665
Armin Töpfer retweeted
ibo
@ibothinksSep 25
how it feels to use unlimited opus 5.5 before anthropic nerfs it in 2 days
13
27
3
865
37,507
There's no way you can exhaust Claude MAX with Opus 5.5 without doing really stupid things.
1
165
Please don't nerf Opus 5.5!
92
Just ran out of 67% weekly limit left in one hour. Something is broken. @thsottiaux 01a08708-e18f-74f0-8af0-258024bcd427
130
I'll create the Half-Life 2 of bioinformatics.
1
277
Been running Ox-Alpha on a task for more than 12 hours so far that every bioinformatician told me can't be done efficiently. We'll see, we'll see.
1
1
392
Ox-Alpha, inferior GPU kernel developer compared to Fable.
1
2
234
Load more