An ensemble learning approach for modeling the systems biology of drug-induced injury

Abstract

Background Drug-induced liver injury (DILI) is an adverse reaction caused by the intake of drugs of common use that produces liver damage. The impact of DILI is estimated to affect around 20 in 100,000 inhabitants worldwide each year. Despite being one of the main causes of liver failure, the pathophysiology and mechanisms of DILI are poorly understood. In the present study, we developed an ensemble learning approach based on different features (CMap gene expression, chemical structures, drug targets) to predict drugs that might cause DILI and gain a better understanding of the mechanisms linked to the adverse reaction. Results We searched for gene signatures in CMap gene expression data by using two approaches: phenotype-gene associations data from DisGeNET, and a non-parametric test comparing gene expression of DILI-Concern and No-DILI-Concern drugs (as per DILIrank definitions). The average accuracy of the classifiers in both approaches was 69%. We used chemical structures as features, obtaining an accuracy of 65%. The combination of both types of features produced an accuracy around 63%, but improved the independent hold-out test up to 67%. The use of drug-target associations as feature obtained the best accuracy (70%) in the independent hold-out test. Conclusions When using CMap gene expression data, searching for a specific gene signature among the landmark genes improves the quality of the classifiers, but it is still limited by the intrinsic noise of the dataset. When using chemical structures as a feature, the structural diversity of the known DILI-causing drugs hampers the prediction, which is a similar problem as for the use of gene expression information. The combination of both features did not improve the quality of the classifiers but increased the robustness as shown on independent hold-out tests. The use of drug-target associations as feature improved the prediction, specially the specificity, and the results were comparable to previous research studies.

Predicting Drug-Induced Liver Injury Using Machine Learning on a Diverse Set of Predictors

“…The DILI classifications for this challenge also changed from binary to a most, less, ambiguous, and no-DILI concern which is in line with the FDA DILIrank dataset. Predictive model rates from multiple distinct approaches to this challenge in 2019 often yielded similar accuracy results around 0.70 (Aguirre-Plans et al, 2021;Lesiński et al, 2021;Liu et al, 2021). While it is difficult to make a direct comparison across the years of these challenges considering how the fundamental elements of predictive modeling, such as the data sources and classifications, have changed, the goal of the challenge has remained the same in modeling the risk of a drug to lead to liver injury in patients.…”

“…This approach produced models with an accuracy of 75.9% that were also able to correctly identify targets associated with the mechanism of action and toxicity of nonsteroidal antiinflammatory drugs, a class of drugs commonly associated with DILI. Aguirre et al utilized the widest array of predictive data, including L1000 CMap expression, drug-target associations, structural data, phenotype-associated gene signatures, proteinprotein interactions, and drug targets data (Aguirre-Plans et al, 2021). Their models' accuracy remained comparable to other study results at 70%, but they also identified structural dissimilarities within the DILI risk labels used.…”

Artificial genetic polymers against human pathologies

“…Applications of AGPs will be further improved through the use of special dedicated software [174], detection of PNAs using nanopore platforms [175], and with rational improvements in reducing side effects such as hepatotoxicity [176]. Machine learning algorithms partially help alleviate this problem by predicting possible side effects of experimental drugs, however, a lot of work needs to be done in this direction [177][178][179] In terms of exciting perspectives that may open sometimes for oncological applications of AGPs, arming of oncolytic viruses with transgenes encoding AGP-synthesizing polymerases inside the cancer cells would be of great interest, since such an approach may greatly boost the action of cytotoxic drugs.…”

An Algorithm Framework for Drug-Induced Liver Injury Prediction Based on Genetic Algorithm and Ensemble Learning

“…The Graph neural network has been proved to be the most advanced model in many fields, such as molecular attribute prediction [ 13 , 14 ] and has also achieved high prediction accuracy in DILI prediction [ 15 , 16 ]. At the same time, ML models based on toxicogenomics have also achieved high accuracy [ 2 ]; however, their accuracy is highly dependent on experimental data, and the experimental accuracy on large datasets is not satisfactory (the best-obtained accuracy is 0.7) [ 17 ].…”