Torfinn Nome
@torfinnnome
Followers
438
Following
5K
Media
15
Statuses
508
Bioinformatics, tech, HPC, University of Oslo, Norway, Obviously moved on to @[email protected]
Joined January 2008
Come see what we've been cooking up for you! @nanopore and community @nextflowio pipelines wrapped up in a GUI suitable for everyone: from baby bioinformaticians to Gentoo gurus! We're on Linux, Windows and even Mac! IT'S ALSO FREE so give it a try today & tell us what you think!
Installing and using bioinformatics software can be difficult for the uninitiated. EPI2ME Labs allows the analysis of anything, anywhere, by anyone.
1
12
48
The right way to evaluate chatGPT isn't by focusing on its current capabilities or limitations, but by the pace of improvement since GPT-3. I think at this rate we're likely to have intelligent AI research assistants <10y, which will drastically increase the pace of science.
5
6
35
For Christmas, I made you 25 days of free programming puzzles. I hope you like them. 💕 🚨 We're less than two hours away from this year's #AdventOfCode 2022! 🚨 PS: How'd you first hear about AoC? I don't exactly advertise.
108
247
2K
Islamabad i dag, og ikke et pip i norske nyheter! Hele Rawalpindi og Islamabad er helt stengt ned. Landet står på randen for konkurs. Foruten at dette har skikkelige geopolitiske konsekvenser, er dette direkte anliggende en betydelig andel norske innbyggere. Men ikke et pip.
9
30
200
As of next year “Next Generation” sequencing will have been around (2005-2023) as long as the break between the original Star Trek and The Next Generation (1969-1987). It’s ok to stop calling it that
19
51
303
Galactica is basically GPT-3 for science. It can write whitepapers, reviews, wikipedia pages and code. It knows how to cite and how to write equations. It's kind of big deal 1/ 🧵
🪐 Introducing Galactica. A large language model for science. Can summarize academic literature, solve math problems, generate Wiki articles, write scientific code, annotate molecules and proteins, and more. Explore and get weights: https://t.co/jKEP8S7Yfl
12
253
1K
Breaking: Norwegian authorities named the real identity of the fake "Brazilian researcher": his name is Mikhail Mikhushin, born 1978. This announcement is in line with what we, together with our investigative partners, had already discovered, and I will add some more details:
Wow. Norway arrested a "Brazilian" university researcher at University of Tromso, who studied "hybrid threats". The catch? He's a Russian spy, not a Brazilian professor. More to come soon. https://t.co/cHwRZ22QBA
82
2K
8K
James Lovelock, who created the Gaia hypothesis, has died on his 103rd birthday. One of the most influential scientists of our time, he worked for the British government during the second world war and later for NASA on the Mars Viking mission. https://t.co/mje1HzKDAd
16
240
666
No need to create containers as tool author. For every #bioconda package your get a @BioContainers (@SingularityApp, @Docker) automatically. Reproducible and transparent! Get all 60.000 containers via the #usegalaxy @cvmfs - a read only http filesystem.
Poll concluded with 46% of responders having used ≥2 bioinfo tools via containers in the past 3 months. On one hand, docker/etc is gaining ground in this field. On the other hand, >50% of researchers are not used to containers. Worth considering when distributing tools.
0
10
36
If you're doing a lot of remote #bioinformatics (on an HPC cluster or cloud node) I can't recommend @VisiData enough. It's a lovely terminal-based "spreadsheet" viewer for Excel/CSV/parquet etc, which you can use to poke at your tabular data without needing to open R or Python.
3
39
236
Two job positions are available in our group @CIGENE_NMBU. Come and work in beautiful Norway with us using LONG READ data to build STRUCTURAL VARIANT aware genomes in cattle and pigs! https://t.co/gb4Tzbhoic
0
25
19
The rapid @nanopore improvements means that I had to update my slides from last years. R10.3 was roughly were R9.4.1 is today and well R10.4 simplex is now ~99% and for duplex things start to get really crazy. A few more iterations and people will stop talking about accuracy?
7
53
121
Introducing DeepConsensus, a new method for correcting sequencing errors. By improving the quality of sequence data, researchers may be able to assemble more difficult parts of the genome, further improve accuracy and reduce costs in the future @GoogleAI
github.com
DeepConsensus uses gap-aware sequence transformers to correct errors in Pacific Biosciences (PacBio) Circular Consensus Sequencing (CCS) data. - google/deepconsensus
2
29
84
To help those looking at Ensembl's rapid release of the new Atlantic salmon assembly (Ssal3.1) we have available an annotation table matching the new Ensembl ids to the previous assembly annotation: https://t.co/o5oEpW8gwc This is to help before a proper annotation comes later.
gitlab.com
GitLab.com
0
5
6
Demonstrating how fast (both implementation time and runtime) SLOW5 format can be: spent around 15 minutes to get slow5 working on @haowen_zhang's sigmap tool. Result: mapping 80k reads that took around 2 hours with FAST5, now takes only 5 minutes with SLOW5! That is >100X faster
1
21
73
I am thrilled to present mm2-fast: an accelerated version of Minimap2 that achieves up to 3.5x speedup on CPUs while maintaining identical output. Code: https://t.co/yjjNhF41EK
#genomics #longreads #Minimap2 #HPC @Saurabh_Kalikar @chirgjain @wasim_galaxy @lh3lh3
Accelerating long-read analysis on modern CPUs https://t.co/0ZNrYUssm7
#bioRxiv
3
84
262
Ben and I have released GPT-J, 6B JAX-based Transformer LM 🥳 - Performs on par with 6.7B GPT-3 - Performs better and decodes faster than GPT-Neo - repo + colab + free web demo article: https://t.co/a3uDbYtHwg repo: https://t.co/RL4vshKfXg
26
237
1K