Using machine learning classifiers to identify the critical proteins in Down syndrome

Abstract

Pharmacotherapies of intellectual disability (ID) are largely unknown as the abnormalities at the complex molecular level which causes ID are difficult to understand. Down syndrome (DS) which is the prevalent cause of ID and caused by an extra copy of the human chromosome21 (Hsa21) has been investigated on protein levels by using the Ts65Dn mouse model of DS which are orthologs of %50 of Hsa21 classical protein coding genes. Recent works have applied the classification methods to understand critical factors in DS as it is believed that the problem was naturally related to classification problem since the determination of proteins discriminatory between classes of mice was required. In this study we apply forward feature selection method to identify correlated proteins and their interactions in DS. After identification we report supervised learning model of expression levels of selected proteins in order to understand the critical proteins for diagnosing and explaining DS. The proposed technique depicts optimum classification results achieved by optimizing parameters with grid search. When compared with the former work our classification results give higher accuracy. © 2018 Association for Computing Machinery.