A Machine Learning Protocol for Predicting Protein Infrared Spectra

• DOI: 10.1021/jacs.0c06530
• Publication Date: 2020-10-31T03:39:15Z
• Authors: Sheng Ye; Kai Zhong; Jinxiao Zhang; Wei Hu; Jonathan D. Hirst; Guozhen Zhang; Shaul Mukamel; Jun Jiang
• Keywords: Biomolecule; Transferability; Folding (DSP implementation); Quantum chemical

Infrared (IR) absorption provides important chemical fingerprints of biomolecules. Protein secondary structure determination from IR spectra is tedious since its theoretical interpretation requires repeated expensive quantum-mechanical calculations in a fluctuating environment. Herein we present novel machine learning protocol that uses few key structural descriptors to rapidly predict amide I various proteins and agrees well with experiment. Its transferability enabled us distinguish protein structures, probe atomic variations temperature, monitor folding. This approach offers cost-effective tool model the relationship between their biological/chemical properties.