Chapter 15 - Machine learning in translational bioinformatics

Abstract

Machine learning (ML) has become an important tool to decipher large and complex biologic datasets. Because of this, it has become the go-to tool. It has been used to deal with the wide array of bioinformatics applications. As times have progressed, the abundance of the biologic datasets has increased, which has led to the need for novel approaches, as opposed to classical conventional methods. Translational bioinformatics (TBI) is a relatively new field that integrates biomedical data science and informatics. It deals with problems ranging from basic biomedical research to clinical practices. The abundance of high-throughput data and advancement in storage and computational power has led to the transformation of biologic information into a data-rich establishment to deal with the exponential growth of high-throughput biologic data, which mainly comprises genomics, proteomics, and transcriptomics. The translational technologies have to keep up with the large amount of data generated from the aforesaid avenues, which in turn will enable the medical practitioner to customize decisions considering individual patients. ML-based techniques have been employed to filter the dataset and to find out all the important features. Thereafter, dimensionality reduction is carried out, and then finally the patterns are found, where these patterns then can be used to infer a hypothesis (pipeline for ML-based analysis). This chapter details the basic background of TBI and ML. Apart from that, it enlists the applications of ML and ML-based methods in various associated and subfields of TBI and how it is used and can be used for better analysis and further research in different avenues, and finally, it discusses the limitations of ML in TBI.