@OriolVinyalsML The success of #AlphaFold2 at #CASP14 is a testament to how far AI has come in aiding scientific discovery. A giant leap for biology!

Protein structure prediction by AlphaFold2: are attention and symmetries all you need? >> ow.ly/9xFC50NGboh @ActaCrystD @IUCr #OpenAccess #AlphaFold2 #ProteinStructurePrediction #CASP14

🗣@Alfons_Valencia "After what we saw as assessors of #CASP14 it was easy to predict that this will happen sooner than later. Congratulations - your work is driving and transforming our community!"

Fundación BBVA @FundacionBBVA

a year ago

🗣@Alfons_Valencia "After what we saw as assessors of #CASP14 it was easy to predict that this will happen sooner than later. Congratulations - your work is driving and transforming our community!"

0 16 29 7K 0

Download Image

#AlphaFold es el sistema de Inteligencia Artificial que en 2018 y 2020 logró determinar con gran precisión el plegamiento de proteínas a partir de una secuencia de aminoácidos dada, en el CASP (Critical Assessment of Protein Structure Prediction). #CASP13 #CASP14

Protein folding competition #CASP15 has followed the same pattern observed in computer vision back in 2012, revolutionized by Deep Learning. Following dominance of #AlphaFold2 in #CASP14, all top methods this year build on top of AF2.

Slides of the different #CASP15 presentations available for download. Slide below shows how #CASP14 (black line) was truly the renaissance moment of protein folding

Sergey Ovchinnikov 🇺🇦 @sokrypton

a year ago

Slides of the different #CASP15 presentations available for download. Slide below shows how #CASP14 (black line) was truly the renaissance moment of protein folding https://t.co/IgLK9UEEoe

2 25 122 0 17

Download Image

The amount of progress from #CASP14 to #CASP15... 🤯

So, the significant assembly pred. improvement we witness in #CASP15 is a reflection of the tertiary structure pred. revolution observed in #CASP14. Even so, smart tweaks performed by several academic and industry groups was needed to push the multimeric modeling limits further.

Over these targets, community produced extremely good results, when we consider the interface patch (IPS) and interface contact (ICS) scores. In #CASP15, for almost 50% of the cases, there is at least one high accuracy model generated (ICS>=0.8), which was only 7% in #CASP14.

Compared to #CASP14, in #CASP15, we have a significant increase in the number of targets offered and in the number of groups participating the challange:

Congratulations to this year’s #CASP15 winners and participants for their achievements! 🧬 In the two years since #CASP14, we’ve been excited to see the advances being made in predicting protein structures, building on #AlphaFold and bringing in new ideas.

Very excited for the release of this years #CASP15 results for protein structure prediction TOMORROW. After the milestone reached by #AlphaFold2 and #Rosetta two years ago, I think tomorrow will look a bit different then at #CASP14. Especially the multimer-category will be hot!

Protein structure prediction by AlphaFold2: are attention and symmetries all you need? >> ow.ly/elW150Lmxu9 @ActaCrystD @IUCr #OpenAccess #AlphaFold2 #ProteinStructurePrediction #CASP14

And now #GaetanoMontelione @rpi on how #AlphaFold can help us analyse #NMRdata -he's looking at the 3 NMR structures that made it into #CASP14- fab talk that I've actually heard some of before @ccpnmr though not in person but anyway I know the punchline! #nospoilers

Protein structure prediction by AlphaFold2: are attention and symmetries all you need? >> ow.ly/e7l350Jhzc8 @ActaCrystD @IUCr #OpenAccess #AlphaFold2 #ProteinStructurePrediction #CASP14

Why #alphafold as in #casp13 was not improbable and a new strategy was needed - #alphafold2 origin and #casp14 success. (It was a privilege to be part of CAsP assessors to see this happening form the first row- certainly “a moment” in modern science - thanks CASP organizers!!

#AlphaFold 2.0, released in 2020, has been called the gold standard for predicting protein structures and is the first computational method to achieve near experimental accuracy with a Global Distance Score averaging 92.4 in #CASP14- predicting structures to the nearest atom 2/6

Credit is also due to the many, many computational structural biologists that have worked over the past several decades to develop and improve algorithms to predict protein 3D structure given a linear amino acid sequence, facilitated by competitions such as #CASP14.