Classification models distinguish functional and trafficking effects of KCNQ1 variants to enhance variant interpretation
Authors:
- Ana C. Chang-GonzalezACAna C. Chang-Gonzalez
- Eric W. BellEBEric W. Bell
- Carlos G. VanoyeCVCarlos G. Vanoye
- Eduardo GuadarramaEGEduardo Guadarrama
- Reshma R. DesaiRDReshma R. Desai
- Jean-Marc DeKeyserJDJean-Marc DeKeyser
- Kathryn R. ButcherKBKathryn R. Butcher
- Thomas ScottTSThomas Scott
- Charles R. SandersCSCharles R. Sanders
- Alfred L. GeorgeAGAlfred L. George
- Kaitlyn V. LedwitchKLKaitlyn V. Ledwitch
- Jens MeilerJMJens Meiler
Preprint
Abstract
Missense mutations compromise protein fitness by altering stability and function, which can lead to various clinical disease states. The potassium ion channel KCNQ1 underlies the majority of congenital long QT syndrome (LQTS) cases, one of the most common genetic arrhythmia syndromes. During genetic testing for LQTS, variants of uncertain significance (VUS) confound diagnosis and clinical management. KCNQ1 protein fitness metrics enable mechanistic classification of variants, directly informing the molecular basis for dysfunction and providing clinical interpretation of variants linked to LQTS and other channelopathies. We developed structure-aware random forest classifier models to predict seven metrics of KCNQ1 fitness, four functional electrophysiology measurements (peak current density, voltage-dependence, gating kinetics), and three trafficking values measured by flow cytometry. Our trained models outperformed AlphaMissense in predicting protein fitness, enhancing interpretation of ClinVar VUS and variants classified as ambiguous by AlphaMissense. We demonstrate the classifiers distinguish benign and pathogenic variants from ClinVar and gnomAD and identify systematic patterns of dysfunction and mistrafficking along the functionally critical S4 helix. Our method advances variant effect prediction with a mechanistic classifier that reliably links missense mutations in KCNQ1 to their specific disease-causing mechanisms. As a resource for precision medicine approaches for LQTS or other KCNQ1 channelopathies, we provide the predictions and scores for all KCNQ1 missense variants across the structured region of the protein.Keywords
Funding Information
- Deutsche Forschungsgemeinschaft (DFG, German Research Foundation (SFB1423, project number 421152132)
This publication has 61 references indexed in Scilit:
- Critical assessment of missense variant effect predictors on disease-relevant variant dataHuman Genetics, 2025
- Structure-informed protein language models are robust predictors for variant effectsHuman Genetics, 2024
- Variant effect predictors: a systematic review and practical guideHuman Genetics, 2024
- The membrane protein KCNQ1 potassium ion channel: Functional diversity and current structural insightsBiochimica et Biophysica Acta (BBA) - Biomembranes, 2020
- Exploring the use of molecular dynamics in assessing protein variants for phenotypic alterationsHuman Mutation, 2019
- What went wrong with variant effect predictor performance for the PCM1 challengeHuman Mutation, 2019
- Assessment of methods for predicting the effects of PTEN and TPMT protein variantsHuman Mutation, 2019
- Membrane Protein Structure, Function, and Dynamics: a Perspective from Experiments and TheoryThe Journal of Membrane Biology, 2015
- The Evaluation of Tools Used to Predict the Impact of Missense Variants Is Hindered by Two Types of CircularityHuman Mutation, 2015
- Assessing predictive accuracy: How to compare brier scoresJournal of Clinical Epidemiology, 1991