MutaBind2: Predicting the Impacts of Single and Multiple Mutations on Protein-Protein Interactions

Abstract

Missense mutations may affect proteostasis by destabilizing or over-stabilizing protein complexes and changing the pathway flux. Predicting the effects of stabilizing mutations on protein-protein interactions is notoriously difficult because existing experimental sets are skewed toward mutations reducing protein-protein binding affinity and many computational methods fail to correctly evaluate their effects. To address this issue, we developed a method MutaBind2, which estimates the impacts of single as well as multiple mutations on protein-protein interactions. MutaBind2 employs only seven features, and the most important of them describe interactions of proteins with the solvent, evolutionary conservation of the site, and thermodynamic stability of the complex and each monomer. This approach shows a distinct improvement especially in evaluating the effects of mutations increasing binding affinity. MutaBind2 can be used for finding disease driver mutations, designing stable protein complexes, and discovering new protein-protein interaction inhibitors.

SAAMBE-3D: Predicting Effect of Mutations on Protein–Protein Interactions

“…For example, we noticed that our old method, SAAMBE, which achieved PCC of 0.62 on the SKEMPI v1.1 database, performed really badly on SKEMPI v2.0, PCC = 0.45. Similar observations were made by other researchers [50]. Therefore, in the next paragraph, we present comparisons and benchmarking for (a) methods that are trained on the same dataset and (b) on blind tests set of data not used in the training.…”

“…We believe this is the only way we can make a fair comparison because there is no information available about the training and test sets used for mCSM-PPI2/MutaBind2 methods. The performance of SAAMBE-3D and mCSM-PPI2/MutaBind2 is compared on 3753/3073 single mutation experimental ∆∆G from SKEMPI v2.0 and presented in Figure 3a We compared the prediction performance of SAAMBE-3D with MCSM-PPI2 [49], and MutaBind2 [50], which are the only two available methods in the literature, trained against SKEMPI v2.0 [1]. We adopted similar purging procedures as the above references, resulting in two datasets: dataset A (which is dataset-1) for comparison with mCSM-PPI2, and dataset B made of 3073 mutations from 257 proteins for comparison with MutaBind2.…”

“…We compared PCC of experimental and predicted ΔΔG for available methods. The methods considered for comparison here are iSee [48], mcSM [33], BindProfX [47], FoldX [41], mCSM-PPI2 [49] and MutaBind2 [50]. The PCC values for this dataset using iSee, mCSM, BindProfX and FoldX are taken from Geng et al's paper [48].…”

“…In this way, 56 single mutations were selected from three complexes (PDB ID: 1IAR, 1VFB and 1YCR). SAAMBE-3D was compared with mCSM-PPI2 [49] and MutaBind2 [50] and the correlation between experimental and predicted ΔΔG for the 56 mutations is presented in Figure 6. We could not calculate ΔΔG using BindProfX as two complexes contain more than two chains and BindProfX can only predict ΔΔG for dimer protein complexes.…”

“…Additionally, MutaBind is specially trained on the protease-inhibitor complexes of SKEMPI 1.1, accomplishing PCC of 0.76. Recent development of MutaBind, MutaBind2 [50], resulted in PCC of 0.82 in benchmarking test on SKEMPI v2.0 database. Moreover, MutaBind2 can predict binding free energy change due to multiple mutations.…”