August 20, 2018Open Access

Predicting changes to INa from missense mutations in human SCN5A

Q: What is the clinical evidence from this study?

Study design: Other. Population: SCN5A missense mutations (n=243). Intervention: Machine-learning models using variant features vs. Zero-R baseline / Existing pathogenicity predictors. Primary outcome: Prediction of complete loss of INa current (zero current) (AUC 0.785).

Key Result

Machine-learning methods predicted complete loss of INa current from SCN5A missense mutations with an accuracy of 91.4% and an AUC of 0.785, outperforming existing pathogenicity predictors.

PICO

Population

Dataset of 243 unique non-synonymous missense mutations in the human SCN5A gene with available cellular electrophysiology data used to train machine-learning models.

Exposure / Comparator

Machine-learning models using variant features vs Zero-R baseline / Existing pathogenicity predictors

Primary Outcome

Prediction of complete loss of INa current (zero current) — AUC 0.785

Main Result

Effect estimate: AUC 0.785

Absolute Event Rate: 91.4% vs 87.7%

Limitations

Imbalanced datasets biased towards the majority group (most reported variants have an effect).
Inconsistencies in EP data reports across different studies.
Lack of precise information about the type of alpha-subunit used in many papers.
Need for more descriptive features to capture differences in physical properties before and after amino-acid substitution.

Abstract

Abstract Mutations in SCN5A can alter the cardiac sodium current I Na and increase the risk of potentially lethal conditions such as Brugada and long-QT syndromes. The relation between mutations and their clinical phenotypes is complex, and systems to predict clinical severity of unclassified SCN5A variants perform poorly. We investigated if instead we could predict changes to I Na , leaving the link from I Na to clinical phenotype for mechanistic simulation studies. An exhaustive list of nonsynonymous missense mutations and resulting changes to I Na was compiled. We then applied machine-learning methods to this dataset, and found that changes to I Na could be predicted with higher sensitivity and specificity than most existing predictors of clinical significance. The substituted residues’ location on the protein correlated with channel function and strongly contributed to predictions, while conservedness and physico-chemical properties did not. However, predictions were not sufficiently accurate to form a basis for mechanistic studies. These results show that changes to I Na , the mechanism through which SCN5A mutations create cardiac risk, are already difficult to predict using purely in-silico methods. This partly explains the limited success of systems to predict clinical significance of SCN5A variants, and underscores the need for functional studies of I Na in risk assessment.

Bookmark

View Full Paper

Bookmark

View Full Paper

Cite This Study

Clerx et al. (Mon,) conducted a other in SCN5A missense mutations (n=243). Machine-learning models using variant features vs. Zero-R baseline / Existing pathogenicity predictors was evaluated on Prediction of complete loss of INa current (zero current) (AUC 0.785). Machine-learning methods predicted complete loss of INa current from SCN5A missense mutations with an accuracy of 91.4% and an AUC of 0.785, outperforming existing pathogenicity predictors.

synapsesocial.com/papers/6a21a1f9153b2036cbf1e956 https://doi.org/https://doi.org/10.1038/s41598-018-30577-5