Self-consistency test reveals systematic bias in programs for prediction change of stability upon mutation

Usmanova, Dinara R.; Bogatyreva, Natalya S.; Bernad, Joan Ariño; Eremina, Aleksandra A; Gorshkova, A. A.; Kanevskiy, German M; Lonishin, Lyubov R; Meister, Alexander V; Yakupova, Alisa G; Kondrashov, Fyodor A.; Ivankov, Dmitry N.

doi:10.1093/bioinformatics/bty340

Cited by 76 publications

(81 citation statements)

References 30 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The training data sets available so far with experimentally determined protein stability changes are enriched with destabilizing mutations (16,48). Thus, the vast majority of predictors that did not consider the unbalance of training dataset showed a better performance for predicting destabilizing than stabilizing mutations (49,50). A recent study constructed a balanced data set with an equal number of destabilizing and stabilizing mutations and was used to assess the performance of 14 methods (51).…”

Section: Introductionmentioning

confidence: 99%

PremPS: Predicting the Effects of Single Mutations on Protein Stability

Chen¹,

Lu²,

Zhang³

et al. 2020

Preprint

View full text Add to dashboard Cite

Protein stability is related to its functional activities, and effect on stability or misfolding could be one of the major disease-causing mechanisms of missense mutations. Here we developed a novel machine learning computational method PremPS, which predicts the effects of single mutations on protein stability by calculating the changes in unfolding Gibbs free energy. PremPS uses only ten evolutionary-and structure-based features and is parameterized on five thousand mutations.Our approach outperforms previous methods and shows a considerable improvement in estimating the effects of mutations increasing protein stability. In addition, PremPS presents an outstanding performance in predicting the pathogenicity of missense mutations using an experimental dataset composed of two thousand non-neutral and neutral mutations. PremPS can be applied to many tasks, including finding functionally important variants, revealing the molecular mechanisms of functional influences and protein design. It is freely available at https://lilab.jysw.suda.edu.cn/research/PremPS/. Key Points:• Considerable improvement in estimating the effects of mutations increasing protein stability;• Comprehensive comparison with other 25 computational methods on different test sets;• An outstanding performance in predicting the pathogenicity of missense mutations;• PremPS employs only ten distinct features belonging to six categories, and the most important feature describes evolutionary conservation of the site;• The webserver allows to do large-scale mutational scanning and takes about ten minutes to perform calculations for one thousand mutations from a normal size protein.

show abstract

Section: Introductionmentioning

confidence: 99%

PremPS: Predicting the Effects of Single Mutations on Protein Stability

Chen¹,

Lu²,

Zhang³

et al. 2020

Preprint

View full text Add to dashboard Cite

show abstract

“…In contrast, FoldX predicted that 83.2% of pathogenic variants are destabilizing. As already demonstrated in previous studies, the bias is likely because FoldX was parameterized on an experimental ∆∆G data set dominated by destabilizing mutations (Pucci et al, 2018;Thiltgen and Goldstein, 2012;Usmanova et al, 2018).…”

Section: ∆∆G Landscape Of Clinvar Missense Variantsmentioning

confidence: 89%

“…A well-performing, "self-consistent" method for predicting ∆∆ s would not only give accurate ∆∆ predictions for the direct mutations, but also for the reverse mutations. The self-consistency requirement has been largely ignored by previously developed ∆∆ predictors (Pucci et al, 2018;Thiltgen and Goldstein, 2012;Usmanova et al, 2018).…”

Section: Thermodynamics Of Direct and Reverse Mutationsmentioning

confidence: 99%

“…For example, physics-based methods are computationally demanding and low-throughput; these challenges have largely prevented them from being applied to large-scale protein engineering and variant interpretation tasks. In addition, several studies have highlighted significant bias in the predictions of machine learning-based methods; they tend to predict mutations as destabilizing more often than stabilizing (Fariselli et al, 2015;Pucci et al, 2018;Thiltgen and Goldstein, 2012;Usmanova et al, 2018). The main source of this bias likely comes from the facts that the training sets are dominated by experimentderived destabilizing mutations and that machine learning methods are prone to overfitting to training sets (Fang, 2019;Pucci et al, 2018).…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Predicting changes in protein thermodynamic stability upon point mutation with deep 3D convolutional neural networks

Yang

Capra

et al. 2020

Preprint

View full text Add to dashboard Cite

Predicting mutation-induced changes in protein thermodynamic stability (∆∆G) is of great interest in protein engineering, variant interpretation, and drug discovery. We introduce ThermoNet, a deep, 3D-convolutional neural network designed for structure-based prediction of the change in protein thermostability upon point mutation. To naturally leverage the image-processing power inherent in convolutional neural networks, we treat protein structures as if they were multi-channel 3D images. In particular, the inputs to ThermoNet are multi-channel voxel grids based on biophysical properties derived from raw atom coordinates; thus, the interpretability of its resulting predictions is improved compared to machine-learning approaches that take heterogeneous representations of predictive features as inputs. ThermoNet is trained with a data set balanced with direct and reverse mutations generated by symmetry-based data augmentation. It demonstrates improved performance compared to fifteen previously developed computational methods on a widely used blind test set. Unlike all other methods that exhibit a strong bias towards predicting destabilization, ThermoNet accurately predicts the effects of both stabilizing and destabilizing mutations. Finally, we demonstrate the practical utility of ThermoNet in predicting the ∆∆G landscape for two clinically relevant proteins, p53 and myoglobin, and all ClinVar missense variants. These predictions indicate that 85.1% of benign variants have effects on protein stability that are in the range expected to be neutral and that pathogenic variants are almost equally likely to be destabilizing as stabilizing (56.6% vs 43.4%, respectively). Overall, our results suggest that 3D convolutional neural networks can model the complex, non-linear interactions perturbed by mutations, directly from biophysical properties of atoms.

show abstract

“…However, mutations within each subunit of the RNAP complex, and primarily the rifampin binding  subunit, have clinical implications and influence rifampin-resistance outcomes in mycobacterial diseases (Comas et al, 2012). The performance of various structural, sequence and NMA based predictors for predicting protein stability changes upon mutations vary largely in terms of their accuracy and bias (Usmanova et al, 2018), but offer a quick and a helpful alternative to understanding the association between mutations and resistance phenotypes (Pires et al, 2016b).…”

Section: Discussionmentioning

confidence: 99%

Computational Saturation Mutagenesis to predict structural consequences of systematic mutations on protein stability and rifampin interactions in the β subunit of RNA polymerase inMycobacterium leprae

Vedithi

Rodrigues

Portelli

et al. 2019

Preprint

View full text Add to dashboard Cite

The functional consequences of missense mutations within the  subunit of RNA polymerase in Mycobacterium leprae (M. leprae) contribute to phenotypic rifampin resistance in leprosy. Here we report in-silico saturation mutagenesis of the  subunit of RNA polymerase to all other 19 amino acid types and predicted their impacts on thermodynamic stability, interactions at subunit interfaces,  subunit-RNA and rifampin affinities using state-of-the-art structure, sequence and normal mode analysis methods. A total of 21,394 mutations were analysed, and it was noted that mutations in the conserved residues that line the active-site cleft show largely destabilizing effects, resulting in increased relative solvent accessibility and concomitant decrease in depth of the mutant residues. The mutations at residues S437, G459, H451, P489, K884 and H1035 are identified as extremely detrimental as they induce highly destabilizing effects on stability, nucleic acid and rifampin affinities. Destabilizing effects were predicted for all the experimentally identified rifampin-resistant mutations in M. leprae indicating that this model can be used as a surveillance tool to monitor emerging detrimental mutations conferring rifampin resistance in leprosy.

show abstract

Self-consistency test reveals systematic bias in programs for prediction change of stability upon mutation

Cited by 76 publications

References 30 publications

PremPS: Predicting the Effects of Single Mutations on Protein Stability

PremPS: Predicting the Effects of Single Mutations on Protein Stability

Predicting changes in protein thermodynamic stability upon point mutation with deep 3D convolutional neural networks

Computational Saturation Mutagenesis to predict structural consequences of systematic mutations on protein stability and rifampin interactions in the β subunit of RNA polymerase inMycobacterium leprae

Contact Info

Product

Resources

About