Unravelling and Quantifying the Biophysical– Biochemical Descriptors Governing Protein Thermostability by Machine Learning.

Analysis of protein thermostability is vital in protein science to aid the understanding of evolutionary aspects of organic life as well as in protein engineering for modern day industrial applications. In the present study, supervised machine learning (ML) algorithms are employed to unravel potential patterns behind protein thermostability. This computational analysis conclusively shows inverse gamma turns, VIII turns, and the propensity of cysteine (Cys) to be the most important biophysical–biochemical attributes responsible for protein thermostability. From the propensity analysis of amino acids, polar residues, specifically glutamine (Gln) and serine (Ser), and charged residues, specifically glutamic acid (Glu) and lysine (Lys), are found to favor the enhancement of protein thermostability. The study demonstrates the feasibility of assigning quantifiable descriptors of thermostability which is expected to aid protein engineering. [ABSTRACT FROM AUTHOR]