High-Precision Atomic Charge Prediction for Protein Systems Using Fragment Molecular Orbital Calculation and Machine Learning

• DOI: 10.1021/acs.jcim.0c00273
• Publication Date: 2020-06-04T17:36:07Z
• Authors: Koichiro Kato; Tomohide Masuda; Chiduru Watanabe; Naoki Miyagawa; Hideo Mizouchi; Shumpei Nagase; Kikuko Kamisaka; Kanji Oshima; Satoshi Ono; Hiroshi Ueda; Atsushi Tokuhisa; Ryo Kanada; Masateru Ohta; Mitsunori Ikeguchi; Yasushi Okuno; Kaori Fukuzawa; Teruki Honma
• Keywords: Partial charge; Atomic charge; Fragment molecular orbital; Fragment (logic); Force Field

Here, we have constructed neural network-based models that predict atomic partial charges with high accuracy at low computational cost. The were trained using high-quality data acquired from quantum mechanics calculations the fragment molecular orbital method. We succeeded in obtaining highly accurate for three representative systems of proteins, including one large biomolecule (approx. 2000 atoms). novelty our approach is ability to take into account electronic polarization system, which a system-dependent phenomenon, being important field drug design. Our high-precision are useful prediction and expected be widely applicable structure-based designs such as structural optimization, high-speed docking, dynamics calculations.