BIOINFORMATICS AND GENOMICS IN RARE DISEASE DIAGNOSIS: LEVERAGING AI AND NEXT-GENERATION SEQUENCING FOR IDENTIFYING NOVEL GENETIC DISORDERS

Abstract

These rare genetic diseases are considered complex regarding their origin by exhibiting several phenotypes that signify their being real challenges. Conventional diagnostic approaches which include clinical assessment and karyotyping fail quite often because rare phenotypes are associated with phenotypic overlap and limits of conventional testing. New developments within next-generation sequencing technologies and bioinformatics tools will allow us to significantly improve the uncovering of genetic bases for these RGDs, thus enabling increasingly comprehensive and accurate diagnoses. Artificial intelligence is enabled through data-mining particularly in the selective automated development of machine learning (ML) models in order to revolutionize how people investigate the genomic data in future. Such artificial intelligence processes, including deep learning (DL), random forests (RF) and support vector machines (SVM), are integrated into the triad of identifying genetic variants, prioritizing mutations, and prediction of disease outcomes. Thus, this analysis aids the discovery of these novel mutations through NGS-based means, linking them to specific phenotypes of diseases and thus providing personalized diagnostics of rare diseases. The usage of this technology in RGD diagnosis has been discussed in this review, emphasizing its applications in diseases such as Congenital disorders of glycosylation, a group of metabolic disorders. The review also mentions that while significant progress has been made in terms of clinical successes, there are still challenges such as the quality of the data, ethical concerns pertaining to the usage of the models, and transparency in the models themselves. Thus, this paper advocates for the need for multi-omics data integration and better interpretability of AI models, as well as interdisciplinary approaches.