Björn Wallner

If you want to know how we improved upon today's Nobel Prize in Chemistry. It is out in @NatureComms nature.com/articles/s4146… #NobelPrize2024

In our newest improvment to Alphafold, AFsample2, we show that it possible improve sampling by randomly masking MSA columns, to break contacts. biorxiv.org/content/10.110…

DockQ v2 - We have improved everything... But the score is the same, you can still use the 0.23 threshold 😀 Improved automatic quality measure for protein multimers, nucleic acids, and small molecules biorxiv.org/content/10.110…

Our paper on the integration of experimental data and quaternary templates in AlphaFold is out. It is flexible and can be adapted to different use-cases, such as modeling missing or flexible regions, involving interactions with huge complexes and new partners. Check it out!

@eriklindahl And if your predictions are too good they usually are… AUC=1.0 is a signal that something is wrong.

This unfortunately looks pretty severe. If correct, it means several major studies will need retraction, and it's a reminder about the dangers of relying on machine learning based on large-but-imperfect data sets.

East Sweden — som inte är en cool startup utan Region Östergötland, Linköpings kommun och Norrköpings kommun — var en av Almedalsveckans största arrangörer i pengar räknat: 1,1 miljoner för 10 personer på två dagar.

This is the funniest Linux joke I’ve ever read. He wins 💀💀💀

@nicholas_pearce That is a classic first mistake when going to Tällberg.

We even did some light Google-directed rally driving through the forest which definitely didn’t affect my heart rate at all.

The d3ms group has arrived in Tällberg for our first #sweprot conference.

@bkhackenofcl Tröjdragning var utanför straffområdet. Sänder kommentatorerna från häckens kansli? Otroligt partiska. (Håller inte på något av lagen)

42’ Sadiq görs ner i straffområdet – straff till BK Häcken. BK Häcken | 0–0 | IFK Norrköping

Dagens startelva 🐝 Så här startar vi i dagens kvartsfinal mot IFK Norrköping. Det är härligt fotbollsväder och arenan har öppnat! Klä dig gulsvart och försiktig då det är halt på arenan. Varmt välkomna! Mot semifinal 🚀 #bkhäcken

En av de stora utmaningarna inom #biologi har länge varit att knäcka koden för varför ett protein får en specifik form. Artificiell intelligens har lett till genombrott inom fältet. I Björn Wallners #forskning möts #AI och #bioinformatik. @bjornwallner liu.se/nyhet/ai-hjalp…

⁦@gabriielstuder⁩ QA assessor in #CASP15 reminds us about the length dependence of TMscore. Problematic for long sequences.

@tfgg2 @DeepMind And that was an obvious mistake by the assessors

@DeepMind To be clear, the provided structures are the baseline mentioned by Dan Rigden in his talk, requested by the CASP assessors to help their analysis, and are definitely not a CASP entry.

Pleased to say @DeepMind has just released a code update to #AlphaFold 2. This includes updated model weights trained on newer data, minor changes to the codebase to inference larger proteins and predictions for the #CASP15 targets. Find out more: github.com/deepmind/alpha…

@tfgg2 @DeepMind Obviously, @deepmind didn’t want to take the risk of participating in CASP. These predictions are not blind. Just a set of postCASP generated predictions and should not be compared to the true blind predictions in CASP.

#casp15 Björn claims compute cost identical to conference registration cost.

Delägare #poppels

Trodde inte jag var så lätt påverkad