Predictive Power Estimation Algorithm (PPEA)--a new algorithm to reduce overfitting for genomic biomarker discovery

Transcriptomics in Toxicogenomics, Part III: Data Modelling for Risk Assessment

Transcriptomics data are relevant to address a number of challenges in Toxicogenomics (TGx). After careful planning of exposure conditions and data preprocessing, the TGx data can be used in predictive toxicology, where more advanced modelling techniques are applied. The large volume of molecular profiles produced by omics-based technologies allows the development and application of artificial intelligence (AI) methods in TGx. Indeed, the publicly available omics datasets are constantly increasing together with a plethora of different methods that are made available to facilitate their analysis, interpretation and the generation of accurate and stable predictive models. In this review, we present the state-of-the-art of data modelling applied to transcriptomics data in TGx. We show how the benchmark dose (BMD) analysis can be applied to TGx data. We review read across and adverse outcome pathways (AOP) modelling methodologies. We discuss how network-based approaches can be successfully employed to clarify the mechanism of action (MOA) or specific biomarkers of exposure. We also describe the main AI methodologies applied to TGx data to create predictive classification and regression models and we address current challenges. Finally, we present a short description of deep learning (DL) and data integration methodologies applied in these contexts. Modelling of TGx data represents a valuable tool for more accurate chemical safety assessment. This review is the third part of a three-article series on Transcriptomics in Toxicogenomics.

Classification of Cholestatic and Necrotic Hepatotoxicants Using Transcriptomics on Human Precision-Cut Liver Slices

Human toxicity screening is an important stage in the development of safe drug candidates. Hepatotoxicity is one of the major reasons for the withdrawal of drugs from the market because the liver is the major organ involved in drug metabolism, and it can generate toxic metabolites. There is a need to screen molecules for drug-induced hepatotoxicity in humans at an earlier stage. Transcriptomics is a technique widely used to screen molecules for toxicity and to unravel toxicity mechanisms. To date, the majority of such studies were performed using animals or animal cells, with concomitant difficulty in interpretation due to species differences, or in human hepatoma cell lines or cultured hepatocytes, suffering from the lack of physiological expression of enzymes and transporters and lack of nonparenchymal cells. The aim of this study was to classify known hepatotoxicants on their phenotype of toxicity in humans using gene expression profiles ex vivo in human precision-cut liver slices (PCLS). Hepatotoxicants known to induce either necrosis (n = 5) or cholestasis (n = 5) were used at concentrations inducing low (<30%) and medium (30-50%) cytotoxicity, based on ATP content. Random forest and support vector machine algorithms were used to classify hepatotoxicants using a leave-one-compound-out cross-validation method. Optimized biomarker sets were compared to derive a consensus list of markers. Classification correctly predicted the toxicity phenotype with an accuracy of 70-80%. The classification is slightly better for the low than for the medium cytotoxicity. The consensus list of markers includes endoplasmic reticulum stress genes, such as C2ORF30, DNAJB9, DNAJC12, SRP72, TMED7, and UBA5, and a sodium/bile acid cotransporter (SLC10A7). This study shows that human PCLS are a useful model to predict the phenotype of drug-induced hepatotoxicity. Additional compounds should be included to confirm the consensus list of markers, which could then be used to develop a biomarker PCR-array for hepatotoxicity screening.

'Toxgnostics': an unmet need in cancer medicine

If we were to summarize the rationale that underpins medical oncology in a Latin aphorism, it might be 'veneno ergo sum'; that is, I poison, therefore I am. The burden of chemotherapy-associated toxicity is well recognized, but we have relatively few tools that increase the precision of anticancer drug prescribing. We propose a shift in emphasis from the focussed study of polymorphisms in drug metabolic pathways in small sets of patients to broader agnostic analyses to systematically correlate germline genetic variants with adverse events in large, well-defined cancer populations. Thus, we propose the new science of 'toxgnostics' (that is, the systematic, agnostic study of genetic predictors of toxicity from anticancer therapy).