Artificial intelligence integration into biological sciences: Applications, opportunities and challenges

Abstract

The age of artificial intelligence (AI) has started, and its application across various fields of biology is increasing rapidly. Machine learning (ML) models are commonly used in the field of genomics for the accurate identification of biological signals and the prediction of gene expression from histone modifications, splicing targets, and TF-binding sites in DNA or RNA. Not only that, machine learning (ML) models can directly analyze MS spectra by using peptide spectral libraries. Models such as XL-MSDigger can predict peptide sequences from MS spectra, while DLDN-Bench can perform de novo peptide sequencing. In the field of metabolomics, AI-based models can identify selective biomarkers by directly analyzing LC/MS data. Such platforms include Deep Metabolome and MeltDB 2.0. Additionally, deep learning (DL) models are revolutionizing the field of glycobiology by identifying complex glycans, their interactions with other biomolecules, stereoselectivity, and glycosylation chemistry. Therefore, an overview of all such relevant platforms is presented here. This review also provides information on how AI platforms help drug discovery and disease diagnostics. Moreover, this review emphasizes the responsible use of AI and best practices.