Applications of AlphaFold beyond Protein Structure Prediction

Solving the half-century-old protein structure prediction problem by DeepMind's AlphaFold is certainly one of the greatest breakthroughs in biology in the twenty-first century. This breakthrough paved the way for tackling some previously highly challenging or even infeasible problems in structural biology. In this study, we propose strategies to use AlphaFold to address several fundamental problems: (1) protein engineering by predicting the experimentally measured stability changes using the representations extracted from AlphaFold models; (2) estimating the designability of a given protein structure by combining a protein design method (e.g. ProDCoNN), sequential Monte Carlo, and AlphaFold. The designability of a protein structure is defined as the number of sequences that encode that protein structure.; (3) predicting protein stabilities using natural sequences and designed sequences as training data, and representations extracted from AlphaFold models as input features; and (4) understanding the sequence-structure relationship of proteins by computational mutagenesis and testing the foldability of the mutants by AlphaFold. We found the representations extracted from AlphaFold models can be used to predict the experimentally measured stability changes accurately. For the first time, we have estimated the designability for a few real proteins. For example, the designability of chain A of FLT3 ligand (PDB ID: 1ETE) with 134 residues was estimated as 3.12 ± 2.14E85.

Exploring the Non-Uniform Memory Access Parallel Architecture Applied to the Protein Structure Prediction Problem

Proteins are base molecules present in live organisms. The study of their structures and functions is of considerable importance for many application fields, particularly for the pharmaceutical area. However, predict the structure of a protein is considered a complex problem. As optimizing methods for this problem have high execution time, a parallel algorithm was proposed. However, just employing parallelization is not enough to guarantee the efficient use of the available computational resources. In this work, the proposed PSP optimizer was executed in a system with NUMA architecture. To demonstrate the effects of this architecture on the execution of an algorithm with simple parallel model, experiments were carried. Results shows that the that the improper execution of a parallel algorithm in this architecture may lead to performance loss.