GAINSeq: glaucoma pre-symptomatic detection using machine learning models driven by next-generation sequencing data

Abstract

Congenital glaucoma, a complex and diverse condition, presents considerable difficulties in its identification and categorization. This research used Next Generation Sequencing (NGS) whole-exome data to create a categorization framework using machine learning methods. This study specifically investigated the effectiveness of decision tree, random forests, and support vector classification (SVC) algorithms in distinguishing different glaucoma genotypes. Proposed methodology used a range of genomic characteristics, such as percentage variation, PhyloP scores, and Grantham scores, to comprehensively understand the genetic pathways that contribute to the illness. This investigation showed that Decision Tree and Random Forest algorithms consistently performed better than earlier techniques in identifying congenital glaucoma subtypes. These algorithms demonstrated outstanding accuracy and resilience. The findings highlight the capacity of machine learning methods to reveal complex patterns in NGS data, therefore improving the proposed comprehension of the causes of congenital glaucoma. Moreover, the knowledge obtained from this research shows potential for enhancing the accuracy of diagnoses and developing tailored treatment approaches for afflicted people.