Dataset Information

Predicting pathogenicity of missense variants with weakly supervised regression.

ABSTRACT: Quickly growing genetic variation data of unknown clinical significance demand computational methods that can reliably predict clinical phenotypes and deeply unravel molecular mechanisms. On the platform enabled by the Critical Assessment of Genome Interpretation (CAGI), we develop a novel "weakly supervised" regression (WSR) model that not only predicts precise clinical significance (probability of pathogenicity) from inexact training annotations (class of pathogenicity) but also infers underlying molecular mechanisms in a variant-specific manner. Compared to multiclass logistic regression, a representative multiclass classifier, our kernelized WSR improves the performance for the ENIGMA Challenge set from 0.72 to 0.97 in binary area under the receiver operating characteristic curve (AUC) and from 0.64 to 0.80 in ordinal multiclass AUC. WSR model interpretation and protein structural interpretation reach consensus in corroborating the most probable molecular mechanisms by which some pathogenic BRCA1 variants confer clinical significance, namely metal-binding disruption for p.C44F and p.C47Y, protein-binding disruption for p.M18T, and structure destabilization for p.S1715N.

SUBMITTER: Cao Y

PROVIDER: S-EPMC6744350 | biostudies-literature | 2019 Sep

REPOSITORIES: biostudies-literature

ACCESS DATA

Publications

Predicting pathogenicity of missense variants with weakly supervised regression.

Cao Yue Y Sun Yuanfei Y Karimi Mostafa M Chen Haoran H Moronfoye Oluwaseyi O Shen Yang Y

Human mutation 20190807 9

Quickly growing genetic variation data of unknown clinical significance demand computational methods that can reliably predict clinical phenotypes and deeply unravel molecular mechanisms. On the platform enabled by the Critical Assessment of Genome Interpretation (CAGI), we develop a novel "weakly supervised" regression (WSR) model that not only predicts precise clinical significance (probability of pathogenicity) from inexact training annotations (class of pathogenicity) but also infers underly ...[more]

PMID: 31144781

Similar Datasets

Project description:BackgroundGermline variants of ten keratin genes (K1, K2, K5, K6A, K6B, K9, K10, K14, K16, and K17) have been reported for causing different types of genodermatoses with an autosomal dominant mode of inheritance. Among all the variants of these ten keratin genes, most of them are missense variants. Unlike pathogenic and likely pathogenic variants, understanding the clinical importance of novel missense variants or variants of uncertain significance (VUS) is the biggest challenge for clinicians or medical geneticists. Functional characterization is the only way to understand the clinical association of novel missense variants or VUS but it is time consuming, costly, and depends on the availability of patient's samples. Existing databases report the pathogenic variants of the keratin genes, but never emphasize the systematic effects of these variants on keratin protein structure and genotype-phenotype correlation.ResultsTo address this need, we developed a comprehensive database KVarPredDB, which contains information of all ten keratin genes associated with genodermatoses. We integrated and curated 400 reported pathogenic missense variants as well as 4629 missense VUS. KVarPredDB predicts the pathogenicity of novel missense variants as well as to understand the severity of disease phenotype, based on four criteria; firstly, the difference in physico-chemical properties between the wild type and substituted amino acids; secondly, the loss of inter/intra-chain interactions; thirdly, evolutionary conservation of the wild type amino acids and lastly, the effect of the substituted amino acids in the heptad repeat. Molecular docking simulations based on resolved crystal structures were adopted to predict stability changes and get the binding energy to compare the wild type protein with the mutated one. We use this basic information to determine the structural and functional impact of novel missense variants on the keratin coiled-coil heterodimer. KVarPredDB was built under the integrative web application development framework SSM (SpringBoot, Spring MVC, MyBatis) and implemented in Java, Bootstrap, React-mutation-mapper, MySQL, Tomcat. The website can be accessed through http://bioinfo.zju.edu.cn/KVarPredDB . The genomic variants and analysis results are freely available under the Creative Commons license.ConclusionsKVarPredDB provides an intuitive and user-friendly interface with computational analytical investigation for each missense variant of the keratin genes associated with genodermatoses.

Dataset Information

Predicting pathogenicity of missense variants with weakly supervised regression.

Publications

Predicting pathogenicity of missense variants with weakly supervised regression.

Similar Datasets

OmicsDI is part of the ELIXIR infrastructure

Tweets