Structure-based prediction of BRAF mutation classes using machine-learning approaches

• DOI: 10.1038/s41598-022-16556-x
• Publication Date: 2022-07-22T10:02:45Z
• Authors: Fanny S. Krebs; Christian Britschgi; Sylvain Pradervand; Rita Achermann; Petros Tsantoulis; Simon Haefliger; Andreas Wicki; Olivier Michielin; Vincent Zoete
• Keywords: Proto-Oncogene Proteins B-raf; 0301 basic medicine; 1000 Multidisciplinary; 0303 health sciences; MAP Kinase Signaling System; Science; Q; R; Mutation, Missense; 610 Medicine & health; Article; 3. Good health; Machine Learning; 03 medical and health sciences; 10032 Clinic for Oncology and Hematology; Mutation; Medicine; Humans

Abstract The BRAF kinase is attracting a lot of attention in oncology as alterations its amino acid sequence can constitutively activate the MAP signaling pathway, potentially contributing to malignant transformation cell but at same time rendering it sensitive targeted therapy. Several pathologic variants were grouped three different classes (I, II and III) based on their effects protein activity pathway. Discerning class mutation permits adapt treatment proposed patient. However, this information lacking new experimentally uncharacterized mutations detected patient biopsy. To overcome issue, we developed silico tool machine learning approaches predict potential missense variant. As I only involves Val600, focused III, which are more diverse challenging predict. Using logistic regression model features including structural information, able known with an accuracy 90%. This fast predictive will help oncologists tackle pathogenic propose most appropriate for patients.