Abstract
Computational methods such as machine learning approaches have a strong track record of success in predicting the outcomes of in vitro assays. In contrast, their ability to predict in vivo endpoints is more limited due to the high number of parameters and processes that may influence the outcome. Recent studies have shown that the combination of chemical and biological data can yield better models for in vivo endpoints. The ChemBioSim approach presented in this work aims to enhance the performance of conformal prediction models for in vivo endpoints by combining chemical information with (predicted) bioactivity assay outcomes. Three in vivo toxicological endpoints, capturing genotoxic (MNT), hepatic (DILI), and cardiological (DICC) issues, were selected for this study due to their high relevance for the registration and authorization of new compounds. Since the sparsity of available biological assay data is challenging for predictive modeling, predicted bioactivity descriptors were introduced instead. Thus, a machine learning model for each of the 373 collected biological assays was trained and applied on the compounds of the in vivo toxicity data sets. Besides the chemical descriptors (molecular fingerprints and physicochemical properties), these predicted bioactivities served as descriptors for the models of the three in vivo endpoints. For this study, a workflow based on a conformal prediction framework (a method for confidence estimation) built on random forest models was developed. Furthermore, the most relevant chemical and bioactivity descriptors for each in vivo endpoint were preselected with lasso models. The incorporation of bioactivity descriptors increased the mean F1 scores of the MNT model from 0.61 to 0.70 and for the DICC model from 0.72 to 0.82 while the mean efficiencies increased by roughly 0.10 for both endpoints. In contrast, for the DILI endpoint, no significant improvement in model performance was observed. Besides pure performance improvements, an analysis of the most important bioactivity features allowed detection of novel and less intuitive relationships between the predicted biological assay outcomes used as descriptors and the in vivo endpoints. This study presents how the prediction of in vivo toxicity endpoints can be improved by the incorporation of biological information—which is not necessarily captured by chemical descriptors—in an automated workflow without the need for adding experimental workload for the generation of bioactivity descriptors as predicted outcomes of bioactivity assays were utilized. All bioactivity CP models for deriving the predicted bioactivities, as well as the in vivo toxicity CP models, can be freely downloaded from https://doi.org/10.5281/zenodo.4761225.
Highlights
Modern toxicity testing heavily relies on animal models, which entails ethical concerns, substantial costs, and difficulties in the extrapolation of results to humans.[1]
We investigated if, and to what extent, the consideration of predicted bioactivities can improve the performance of in silico models for the prediction of the in vivo toxicity endpoints MNT, drug-induced liver injury (DILI), and DICC
For training the models for the three in vivo endpoints, we embedded three types of random forest (RF) models in conformal prediction (CP) frameworks: (a) CHEM models based exclusively on chemical descriptors, (b) BIO models based exclusively on bioactivity descriptors, and (c) chemical and bioactivity descriptors (CHEMBIO) models based on the combination of both types of descriptors
Summary
Modern toxicity testing heavily relies on animal models, which entails ethical concerns, substantial costs, and difficulties in the extrapolation of results to humans.[1]. In silico tools for toxicity prediction have evolved into powerful methods that can help to decrease animal testing.[2−4] This is true when applied in tandem with in vitro methods.[5] Machine learning (ML). Models trained on data sets of compounds with known activities for an assay can be used as predictive tools for untested compounds.[6] These models are generally trained on chemical and structural features of compounds with measured activity values.[7] the outcomes of in vivo toxicological assays depend on a number of biological interactions such as the administration, distribution, metabolism, and excretion (ADME) and the interaction with different cell types.[4] The ability of chemical property descriptors to capture these complex interactions and, the predictive power of ML models trained on these molecular representations are limited. By the example of classification models for hit expansion[8,9] and toxicity prediction,[10−13] recent studies have shown that the predictive power of in silico models can be improved by the amalgamation of chemical and biological
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Free) 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call
