Keywords

Machine Learning Algorithms, Mutation databases, substitution mutation, algorithms. amino acid sequence, protein structure, protein structure prediction, amino acid substitution

Abstract

Although reported gene variants in the RET oncogene have been directly associated with multiple endocrine neoplasia type 2 and hereditary medullary thyroid carcinoma, other mutations are classified as variants of uncertain significance (VUS) until the associated clinical phenotype is made clear. Currently, some 46 non-synonymous VUS entries exist in curated archives. In the absence of a gold standard method for predicting phenotype outcomes, this follow up study applies feature selected amino acid physical and chemical properties feeding a Bayes classifier to predict disease association of uncertain gene variants into categories of benign and pathogenic. Algorithm performance and VUS predictions were compared to established phylogenetic based mutation prediction algorithms. Curated outcomes and unpublished RET gene variants with known disease association were used to benchmark predictor performance. Reliable classification of RET uncertain gene variants will augment current clinical information of RET mutations and assist in improving prediction algorithms as knowledge increases.

Original Publication Citation

Crockett DK, Piccolo SR, Ridge PG, Margraf RL, Lyon E, Williams MS, Mitchell JA. “Predicting phenotypic severity of uncertain gene variants in the RET proto-oncogene.” PLoS ONE 2011, 6:3, e18380

Document Type

Peer-Reviewed Article

Publication Date

2011-03-30

Publisher

Public Library of Science

Language

English

College

Life Sciences

Department

Biology

University Standing at Time of Publication

Associate Professor

Download

Share

COinS