Abstract:Abstract Understanding the effects of missense mutations on protein stability is a widely acknowledged significant biological problem. Genomic missense mutations may alter one or more amino acids leading to increased or decreased stability of the encoded proteins. In this study, we propose a novel approach - Protein Stability Prediction with a Gaussian Network Model (PSP-GNM) to study the effect of single amino acid substitutions on protein stability. Specifically, PSP-GNM employs a coarse-grained Gaussian Network Model (GNM) that has interactions between amino acids weighted by the Miyazawa-Jernigan (MJ) statistical potential. We use PSP-GNM to simulate partial unfolding of the wildtype and mutant structures and then, use the difference in energies of the unfolded wildtype and mutant protein structures to estimate the experimentally obtained unfolding free energy change (ΔΔG). We verify the extent of correspondence between the ΔΔG calculated by PSP-GNM and the ΔΔG obtained experimentally using three datasets: 350 forward mutations from 66 proteins, 2298 forward mutations from 126 proteins and 611 forward and reverse mutations from 66 proteins and observe Pearson correlation coefficient (PCC) as high as 0.58 and root mean-squared error (RMSE) as low as 1.24 kcal/mol. The performance is comparable to the existing state of the art methods. Importantly, we do observe an increase in the correlation to 0.73 and decrease in RMSE to 1.07 when considering only those measurements made close to 25°C and neutral pH, suggesting a strong dependence on temperature and pH. PSP-GNM is written in Python and is available as a free downloadable package at https://github.com/sambitmishra0628/PSP-GNM . Author Summary Understanding how genomic missense mutations impact the thermodynamic stability of encoded proteins is important to understand disease etiology. Specifically, mutant proteins are often functionally inactive and underlie numerous genetic and neurodegenerative diseases. A classic example is sickle cell anemia – a single amino acid change significantly affects hemoglobin’s binding affinity for oxygen. To be able to identify mutations that would likely impact the protein function and stability is therefore essential and is the focus of our study. We present an approach that relies on utilizing the intrinsic dynamics of protein structures to predict the effect of single amino acid mutations (point mutations) on protein stability. In our approach, we model proteins as coarse-grained beads (amino acids) and springs (interactions), simulate protein unfolding and identify putative residue-residue contacts that are broken during the unfolding process. We demonstrate that the knowledge of broken contacts and their order is essential in describing the thermodynamic differences between wildtype and mutant proteins. We also highlight the importance of residue-residue interactions at the mutation site in the context of protein stability prediction. Our findings present novel avenues to interpret how genomic mutations may manifest in the encoded proteins.

PMSPcnn: Predicting protein stability changes upon single point mutations with convolutional neural network

Predicting protein thermal stability changes upon single and multi-point mutations via restricted attention subgraph neural network

DeepDDG: Predicting the Stability Change of Protein Point Mutations Using Neural Networks

Assessing computational tools for predicting protein stability changes upon missense mutations using a new dataset

PremPS: Predicting the impact of missense mutations on protein stability

Protein Stability Changes upon Point Mutations Identified with a Gaussian Network Model Simulating Protein Unfolding Behavior

AI Prediction of Structural Stability of Nanoproteins Based on Structures and Residue Properties by Mean Pooled Dual Graph Convolutional Network

PON-Tm: A Sequence-Based Method for Prediction of Missense Mutation Effects on Protein Thermal Stability Changes

DDMut: predicting effects of mutations on protein stability using deep learning

Differentiating stable and unstable protein using convolution neural network and molecular dynamics simulations

STRUM: structure-based prediction of protein stability changes upon single-point mutation

BayeStab: Predicting Effects of Mutations on Protein Stability with Uncertainty Quantification

PROST: AlphaFold2-aware Sequence-Based Predictor to Estimate Protein Stability Changes upon Missense Mutations

mmCSM-PPI: predicting the effects of multiple point mutations on protein–protein interactions

Transfer learning to leverage larger datasets for improved prediction of protein stability changes

A three-state prediction of single point mutations on protein stability changes

High throughput computing to improve efficiency of predicting protein stability change upon mutation

Predicting Pathology of Missense Mutations through Protein-Specific Evolutionary Pattern

Protein stability models fail to capture epistatic interactions of double point mutations

Unbiased Curriculum Learning Enhanced Global-Local Graph Neural Network for Protein Thermodynamic Stability Prediction