Science

New publication 🚨 Better data is all you need for state-of-the-art protein secondary structure prediction! Presenting PS4, the largest open-source dataset for protein single-sequence prediction 🧬 

Achieves SotA q3 & q8 accuracy on CB513 in zero shots by training on PS4 alone, with no MSA input 🦾

Paper: https://www.future-science.com/doi/10.2144/btn-2023-0024

Dataset and eval code:

Building Proof of IRL Human Connection at https://paired.world • Nanotechnology postgrad @ St Catz, Oxford 

Paper: https://www.future-science.com/doi/10.2144/btn-2023-0024

Dataset and eval code: https://github.com/omarperacha/ps4-dataset