Combining In Vivo Data with In Silico Predictions for Modeling Hepatic Steatosis by Using Stratified Bagging and Conformal Prediction.

Sankalp Jain,Ulf Norinder,Sylvia E Escher,Barbara Zdrazil

doi:10.1021/acs.chemrestox.0c00511

Abstract

Hepatic steatosis (fatty liver) is a severe liver disease induced by the excessive accumulation of fatty acids in hepatocytes. In this study, we developed reliable in silico models for predicting hepatic steatosis on the basis of an in vivo data set of 1041 compounds measured in rodent studies with repeated oral exposure. The imbalanced nature of the data set (1:8, with the “steatotic” compounds belonging to the minority class) required the use of meta-classifiers—bagging with stratified under-sampling and Mondrian conformal prediction—on top of the base classifier random forest. One major goal was the investigation of the influence of different descriptor combinations on model performance (tested by predicting an external validation set): physicochemical descriptors (RDKit), ToxPrint features, as well as predictions from in silico nuclear receptor and transporter models. All models based upon descriptor combinations including physicochemical features led to reasonable balanced accuracies (BAs between 0.65 and 0.69 for the respective models). Combining physicochemical features with transporter predictions and further with ToxPrint features gave the best performing model (BAs up to 0.7 and efficiencies of 0.82). Whereas both meta-classifiers proved useful for this highly imbalanced toxicity data set, the conformal prediction framework also guarantees the error level and thus might be favored for future studies in the field of predictive toxicology.

Highlights

Hepatic steatosis (HS; termed “fatty liver”) is a wellknown and often observed condition in the human population and characterized by the accumulation of lipids/fat in the liver.HS can progress to steatohepatitis and irreversible stages of liver disease including fibrosis, cirrhosis, hepatocellular carcinoma, and death
Visualizing training set (TR) vs test set (TS) in a two-dimensional t-distributed stochastic neighbor embedding (t-SNE)[28] plot helped to verify that the TS compounds have been selected in an unbiased way, Figure 1
The generation of reliable machine learning (ML) models for predicting HS based on small molecules tested in vivo in rodents is described

Summary

Introduction

Hepatic steatosis (HS; termed “fatty liver”) is a wellknown and often observed condition in the human population and characterized by the accumulation of lipids/fat in the liver.HS can progress to steatohepatitis and irreversible stages of liver disease including fibrosis, cirrhosis, hepatocellular carcinoma, and death. Hepatic steatosis (HS; termed “fatty liver”) is a wellknown and often observed condition in the human population and characterized by the accumulation of lipids/fat in the liver. NAFLD can result from different exposure conditions such as high-fat diets, exposure to industrial chemicals and environmental pollutants,[1] or pharmaceuticals.[2,3] The progression of NAFLD might be a result of insulin resistance, changes in microbiota, or predisposing genetic variants resulting in a disturbed lipid homeostasis.[4]. The pathological manifestation of HS is characterized by an excessive accumulation of triglycerides in vacuoles in the cytosol of hepatocytes, leading to macrovesicular or microvesicular steatosis. Macrovesicular steatosis usually shows single large vacuoles in the cytoplasm of hepatocytes with nuclear displacement, whereas microvesicular steatosis is characterized by small diffuse lipid droplets.[5]

Objectives

Results

Conclusion