CrysFormer: Protein Structure Prediction via 3d Patterson Maps and Partial Structure Attention
This work addresses the problem of computationally expensive protein structure prediction for researchers in computational biology, offering an incremental improvement by incorporating additional prior knowledge beyond sequence data.
The paper tackles protein structure prediction by introducing CrysFormer, a transformer-based model that uses protein crystallography and partial structure information to predict electron density maps, achieving accurate predictions with a smaller dataset and reduced computation costs.
Determining the structure of a protein has been a decades-long open question. A protein's three-dimensional structure often poses nontrivial computation costs, when classical simulation algorithms are utilized. Advances in the transformer neural network architecture -- such as AlphaFold2 -- achieve significant improvements for this problem, by learning from a large dataset of sequence information and corresponding protein structures. Yet, such methods only focus on sequence information; other available prior knowledge, such as protein crystallography and partial structure of amino acids, could be potentially utilized. To the best of our knowledge, we propose the first transformer-based model that directly utilizes protein crystallography and partial structure information to predict the electron density maps of proteins. Via two new datasets of peptide fragments (2-residue and 15-residue) , we demonstrate our method, dubbed \texttt{CrysFormer}, can achieve accurate predictions, based on a much smaller dataset size and with reduced computation costs.