TY - JOUR
T1 - Prediction of Compound Bioactivities Using Heat-Diffusion Equation
AU - Hidaka, Tadashi
AU - Imamura, Keiko
AU - Hioki, Takeshi
AU - Takagi, Terufumi
AU - Giga, Yoshikazu
AU - Giga, Mi Ho
AU - Nishimura, Yoshiteru
AU - Kawahara, Yoshinobu
AU - Hayashi, Satoru
AU - Niki, Takeshi
AU - Fushimi, Makoto
AU - Inoue, Haruhisa
N1 - Funding Information:
We would like to express our sincere gratitude to all our coworkers and collaborators; to Misako Takemoto, Eri Ejiri, Ayami Onodera, Yuka Hirabayashi, Yumiko Nakagaito, Shigeru Kondo, Hitoshi Nakamura, Hiromitsu Fuse, Keisuke Imamura, Masashi Toyofuku, Yumi Imai, and Sachiko Itono for their experimental support; and to Shinya Yamanaka, Seigo Izumo, Haruhide Kimura, Takanobu Kuroita, Toshimasa Tanaka, Atsushi Nakanishi, Hidetoshi Shimodaira, and Yuichiro Yada for their scientific discussions. This work was funded in part by a grant from the Research Center Network for Realization of Regenerative Medicine from AMED (H.I.), Research Project for Practical Applications of Regenerative Medicine from AMED (H.I.), and grant-in-aid for scientific research (18K18452 to H.I.). Takeda Pharmaceutical Company Limited was the sponsor of this work. Takeda Pharmaceutical Company Limited was paying the salary of T.N. in relation to this work.
Publisher Copyright:
© 2020 The Authors
PY - 2020/12/11
Y1 - 2020/12/11
N2 - Machine learning is expected to improve low throughput and high assay cost in cell-based phenotypic screening. However, it is still a challenge to apply machine learning to achieving sufficiently complex phenotypic screening due to imbalanced datasets, non-linear prediction, and unpredictability of new chemotypes. Here, we developed a prediction model based on the heat-diffusion equation (PM-HDE) to address this issue. The algorithm was verified as feasible for virtual compound screening using biotest data of 946 assay systems registered with PubChem. PM-HDE was then applied to actual screening. Based on supervised learning of the data of about 50,000 compounds from biological phenotypic screening with motor neurons derived from ALS-patient-induced pluripotent stem cells, virtual screening of >1.6 million compounds was implemented. We confirmed that PM-HDE enriched the hit compounds and identified new chemotypes. This prediction model could overcome the inflexibility in machine learning, and our approach could provide a novel platform for drug discovery.
AB - Machine learning is expected to improve low throughput and high assay cost in cell-based phenotypic screening. However, it is still a challenge to apply machine learning to achieving sufficiently complex phenotypic screening due to imbalanced datasets, non-linear prediction, and unpredictability of new chemotypes. Here, we developed a prediction model based on the heat-diffusion equation (PM-HDE) to address this issue. The algorithm was verified as feasible for virtual compound screening using biotest data of 946 assay systems registered with PubChem. PM-HDE was then applied to actual screening. Based on supervised learning of the data of about 50,000 compounds from biological phenotypic screening with motor neurons derived from ALS-patient-induced pluripotent stem cells, virtual screening of >1.6 million compounds was implemented. We confirmed that PM-HDE enriched the hit compounds and identified new chemotypes. This prediction model could overcome the inflexibility in machine learning, and our approach could provide a novel platform for drug discovery.
UR - http://www.scopus.com/inward/record.url?scp=85097095982&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=85097095982&partnerID=8YFLogxK
U2 - 10.1016/j.patter.2020.100140
DO - 10.1016/j.patter.2020.100140
M3 - Article
AN - SCOPUS:85097095982
VL - 1
JO - Patterns
JF - Patterns
SN - 2666-3899
IS - 9
M1 - 100140
ER -