Predicting Prenatal Developmental Toxicity Based On the Combination of Chemical Structures and Biological Data

Abstract

For hazard identification, classification, and labeling purposes, animal testing guidelines are required by law to evaluate the developmental toxicity potential of new and existing chemical products. However, guideline developmental toxicity studies are costly, time-consuming, and require many laboratory animals. Computational modeling has emerged as a promising, animalsparing, and cost-effective method for evaluating the developmental toxicity potential of chemicals, such as endocrine disruptors, without the use of animals. We aimed to develop a predictive and explainable computational model for developmental toxicants. To this end, a comprehensive dataset of 1244 chemicals with developmental toxicity classifications was curated from public repositories and literature sources. Data from 2140 toxicological high-throughput screening assays were extracted from PubChem and the ToxCast program for this dataset and combined with information about 834 chemical fragments to group assays based on their chemical−mechanistic relationships. This effort revealed two assay clusters containing 83 and 76 assays, respectively, with high positive predictive rates for developmental toxicants identified with animal testing guidelines (PPV = 72.4 and 77.3% during cross-validation). These two assay clusters can be used as developmental toxicity models and were applied to predict new chemicals for external validation. This study provides a new strategy for constructing alternative chemical developmental toxicity evaluations that can be replicated for other toxicity modeling studies.

TIRESIA: An eXplainable Artificial Intelligence Platform for Predicting Developmental Toxicity

“…This work provides strategies for predicting the so-called prenatal developmental toxicity based on bioassay clustering, which unveils latent relationships linking biological mechanisms and structural features. 47 In this work, we instead designed and implemented an XAI predictive framework based on a large pool of established developmental toxicity data. We take a step beyond existing literature by considering not only the design of an accurate AI model for developmental toxicity along with a transparent applicability domain (AD) rooted on a density-based hyperdimensional bounding box, but especially, by providing an explainability framework based on SHAP (SHapley Additive ex-Planations) values.…”

Computational model for fetal skeletal defects potentially linked to disruption of retinoic acid signaling

“…From there, we used unsupervised methodology to classify statistical correlations. This led to inferences using only input vectors without referring to known, or labelled, outcomes for predictive toxicology ( Ciallella et al, 2022 ); however, it neglects non-skeletal targets as well as chemicals that did not perturb in vitro bioactivity profiles of the ATRA assays available for this analysis, or non-ATRA pathways, which are important for health-protective inferences for DART testing ( Rajagopal et al, 2022 ). The k-means clustering ( Figure 4 ) findings for the 48 chemicals were highly relevant, and consistent with corresponding heatmap findings ( Figure 5 ).…”

Human-Based New Approach Methodologies in Developmental Toxicity Testing: A Step Ahead from the State of the Art with a Feto–Placental Organ-on-Chip Platform

“…More recently, interesting modeling efforts explored the combination of chemical information (hundreds of chemical fragments) with biological activity extracted from nearly 2000 toxicological high-throughput screening assays extracted from PubChem and ToxCast [ 155 ]. The inputs were used to group assays based on their chemical–mechanistic relationships and identified two clusters where the in vitro assays were enriched with developmental toxic compounds.…”