Dataset Information

Prediction of Compound Bioactivities Using Heat-Diffusion Equation.

ABSTRACT: 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.

SUBMITTER: Hidaka T

PROVIDER: S-EPMC7733880 | biostudies-literature | 2020 Dec

REPOSITORIES: biostudies-literature

ACCESS DATA

Publications

Prediction of Compound Bioactivities Using Heat-Diffusion Equation.

Hidaka Tadashi T Imamura Keiko K Hioki Takeshi T Takagi Terufumi T Giga Yoshikazu Y Giga Mi-Ho MH Nishimura Yoshiteru Y Kawahara Yoshinobu Y Hayashi Satoru S Niki Takeshi T Fushimi Makoto M Inoue Haruhisa H

Patterns (New York, N.Y.) 20201111 9

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 b ...[more]

PMID: 33336198

Similar Datasets

Project description:ContextThe Chronic Kidney Disease Epidemiology Collaboration (CKD-EPI) equation more accurately estimates glomerular filtration rate (GFR) than the Modification of Diet in Renal Disease (MDRD) Study equation using the same variables, especially at higher GFR, but definitive evidence of its risk implications in diverse settings is lacking.ObjectiveTo evaluate risk implications of estimated GFR using the CKD-EPI equation compared with the MDRD Study equation in populations with a broad range of demographic and clinical characteristics.Design, setting, and participantsA meta-analysis of data from 1.1 million adults (aged ≥ 18 years) from 25 general population cohorts, 7 high-risk cohorts (of vascular disease), and 13 CKD cohorts. Data transfer and analyses were conducted between March 2011 and March 2012.Main outcome measuresAll-cause mortality (84,482 deaths from 40 cohorts), cardiovascular mortality (22,176 events from 28 cohorts), and end-stage renal disease (ESRD) (7644 events from 21 cohorts) during 9.4 million person-years of follow-up; the median of mean follow-up time across cohorts was 7.4 years (interquartile range, 4.2-10.5 years).ResultsEstimated GFR was classified into 6 categories (≥90, 60-89, 45-59, 30-44, 15-29, and <15 mL/min/1.73 m(2)) by both equations. Compared with the MDRD Study equation, 24.4% and 0.6% of participants from general population cohorts were reclassified to a higher and lower estimated GFR category, respectively, by the CKD-EPI equation, and the prevalence of CKD stages 3 to 5 (estimated GFR <60 mL/min/1.73 m(2)) was reduced from 8.7% to 6.3%. In estimated GFR of 45 to 59 mL/min/1.73 m(2) by the MDRD Study equation, 34.7% of participants were reclassified to estimated GFR of 60 to 89 mL/min/1.73 m(2) by the CKD-EPI equation and had lower incidence rates (per 1000 person-years) for the outcomes of interest (9.9 vs 34.5 for all-cause mortality, 2.7 vs 13.0 for cardiovascular mortality, and 0.5 vs 0.8 for ESRD) compared with those not reclassified. The corresponding adjusted hazard ratios were 0.80 (95% CI, 0.74-0.86) for all-cause mortality, 0.73 (95% CI, 0.65-0.82) for cardiovascular mortality, and 0.49 (95% CI, 0.27-0.88) for ESRD. Similar findings were observed in other estimated GFR categories by the MDRD Study equation. Net reclassification improvement based on estimated GFR categories was significantly positive for all outcomes (range, 0.06-0.13; all P < .001). Net reclassification improvement was similarly positive in most subgroups defined by age (<65 years and ≥65 years), sex, race/ethnicity (white, Asian, and black), and presence or absence of diabetes and hypertension. The results in the high-risk and CKD cohorts were largely consistent with the general population cohorts.ConclusionThe CKD-EPI equation classified fewer individuals as having CKD and more accurately categorized the risk for mortality and ESRD than did the MDRD Study equation across a broad range of populations.

Project description:Prediction of the rate and extent of drug absorption upon oral dosing needs models that capture the complexities of both the drug molecule and intestinal physiology. We report here the development of a continuous intestinal absorption model based on the convection-diffusion equation. The model includes explicit enterocyte apical membrane and intracellular lipid radial compartments along the length of the intestine. Physiologic functions along length x are built into the model and include velocity, diffusion, surface areas, and pH of the intestine. Also included are expression levels of the intestinal active uptake transporter OATP2B1 and efflux transporter P-gp. Oral dosing of solution as well as solid (with a dissolution function) was modeled for several drugs. The fraction absorbed (FA) and concentration-time (C-t) profiles were predicted and compared with clinical data. Overall, FA was well predicted upon oral (n = 21) or colonic dosing (n = 11), with four outliers. The overall accuracy (prediction of the correct bin) was 81% with outliers and 90% without outliers. Of the nine solution dosing data sets, six drugs were very well predicted with an exposure overlap coefficient (EOC) > 0.9 and predicted Cmax and Tmax values similar to those observed. Of the six solid dose formulations evaluated, the EOC values were > 0.9 for all drugs except budesonide. The observed precipitation of nifedipine at high doses was predicted by the model. Most of the poor predictions were for drugs that are known to be transporter substrates. As proof of concept, incorporating OATP2B1 and P-gp markedly improved the EOC and predicted Cmax and Tmax for fexofenadine. Finally, the continuous intestinal model accurately recapitulated the known relationships between drug absorption and permeability, solubility, and particle size. Together, these results indicate that this preliminary intestinal absorption model offers a simple and straightforward framework to build in complexities such as drug permeability, lipid partitioning, solubility, metabolism, and transport for improved prediction of the rate and extent of drug absorption.

Dataset Information

Prediction of Compound Bioactivities Using Heat-Diffusion Equation.

Publications

Prediction of Compound Bioactivities Using Heat-Diffusion Equation.

Similar Datasets

OmicsDI is part of the ELIXIR infrastructure

Tweets