Computational Prediction

Share to LINE

Predictive toxicology is one of the newly developed and interdisciplinary study for toxicology in the 21st century, which combines knowledge and tools from such as bioinformatics, in silico modeling and systems biology to predict the safety and health effects of chemical substances to biological systems or the environment. These approaches are also considered as non-animal alternative test methods.

At present, many chemical substances do not have detailed toxicological information. Predictive toxicologic tools or database can be applied to preliminary hazard screening and prioritizing for further toxicological tests. These tools and databases can be also used for understanding the toxic mechanism of chemical substances.

TAAT website will keep collecting and introducing predictive analytics tools and databases which are newly developed and commonly used internationally.

ChemDIS is a chemical-disease inference system based on chemical-protein interactions. By integrating the chemical-protein interactions and protein-disease interactions, the diseases associated with a given chemical can be inferred from the chemical-protein-disease relatioinship. ChemDIS provides enrichment analysis tools for identifying chemical-associated Gene Ontology (GO), pathway (KEGG, Reactome and SMPDB) and Disease Ontology (DO and DOlite) based on a hypergeometric distribution. The chemical-protein interactions in humans were retrieved from STITCH database v5.0 and v4.0. ChemDIS is expected to be a useful chemical-disease inference system for assessing potential risks associated with enviromental chemicals.

Go to ChemDIS

ICE provides high-quality curated data and appropriate tools to support development and evaluation of new, revised, and alternative methods. ICE provides free online access to:

  • Curated in vivo and in vitro test data
  • In silico toxicity predictions and chemical property data
  • Reference chemical lists
  • Computational tools for chemical characterization and predicting toxicity

ICE supports:

  • Data integration: brings together available data, including data on formulations
  • Results exploration: enables dynamic, graphical exploration with publication-quality graphics
  • Data analysis: allows characterization of data using online workflows
  • FAIR data access: data are findable, accessible, interoperable, and reusable

ICE data have been reviewed by NICEATM to ensure that they meet specific quality standards and are useful in evaluating a chemical's safety.

ICE updates-ICE Release Notes
Webinar:The Integrated Chemical Environment: Tools and Data to Support Toxicity Assessments
Go to ICE

The OncoLogic™ model is an expert system that mimics the judgment of human experts by following sets of knowledge rules based on studies of how chemicals cause cancer in animals and humans. OncoLogic™ asks for chemical and use information from the user, and following the knowledge rules incorporated into the system, uses the responses to construct an estimation of the carcinogenicity potential of the chemical.

OncoLogic™ - An Expert System to Evaluate the Carcinogenic Potential of Chemicals
Download OncoLogic™ 9.0 Download OncoLogic™ 8.0

The Toolbox is a free software application that supports reproducible and transparent chemical hazard assessment. It offers functionalities for retrieving experimental data, simulating metabolism and profiling properties of chemicals. These information and tools can be used to find structurally and mechanistically defined analogues and chemical categories, which can serve as sources for read-across and trend analysis for data gap filling.

20 September 2021 – An updated version of the QSAR Toolbox 4.5 was released
Go to QSAR toolbox website

To provide robust QSAR/QSPR models for chemical properties of environmental interest that can be used for regulatory purposes, EPA NCCT created the Open Structure-activity/property Relationship App (OPERA). OPERA is a free and open-source/open-data suite of QSAR models providing predictions for physicochemical properties, environmental fate parameters, and toxicity endpoints. All OPERA models were built on curated data and QSAR-ready chemical structures standardized using an open-source workflow

OPERA is an ongoing collaboration between NICEATM and EPA. Recent additions to OPERA include predictions for: Estrogenic activity from the Collaborative Estrogen Receptor Activity Prediction Project (CERAPP), Androgenic activity from the Collaborative Modeling Project for Androgen Receptor Activity (CoMPARA), Acute oral systemic toxicity from the Collaborative Acute Toxicity Modeling Suite (CATMoS), Physicochemical properties such as acid dissociation constant (pKa) and octanol-water dissociation coefficient (logD).

Reference:Open Structure-activity/property Relationship App
Download OPERA

The Pred-Skin application is based on externally predictive QSAR models of skin sensitization. The models were built using the most extensive database containing human, in vivo (LLNA), in chemico (DPRA), and in vitro (KeratinoSens and H-CLAT) data, addressing all the key steps of the skin sensitization adverse outcome pathway (AOP).

PredSkin was developed as a tool to identify putative skin sensitizers. The integrative Naïve Bayes model was generated using the predictions of each QSAR model developed independently for the five skin sensitization assays.


Skin sensitization is an important toxicological endpoint in drug development and regulatory decision making. Chemical sensitizers act as haptens binding to protein molecules to trigger immune responses that could induce allergic contact dermatitis. To facilitate development of AOP-based computational prediction methods, a novel curated database named SkinSensDB has been constructed by manual curation of published literatures.

Go to SkinSensDB

Tox21BodyMap provides a visualization of mapping 971 Tox21/ToxCast assay targets to regions of the human body, and allows users to create hypotheses about what some of the target tissues may be for specific chemicals. The webserver relies upon data on 9,270 chemicals with Tox21 assay activities, information on biological targets of the assays, and an approach that combines expert knowledge with tissue-specific gene expression levels to map assay targets to organ systems, providing an intuitive display of chemical bioactivity patterns. There are a variety of reasons why the mappings in this application could be inaccurate including, but not limited to, incompatibility of the chemical with in vitro test systems (e.g. volatility), activity cliffs due to limits of solubility and the concentration range used in the in vitro assays, absence of bioactivation/metabolism in the in vitro systems, and chemical-specific toxicokinetic/chemical distribution in vivo. This tool is solely a knowledge mapping and data visualization resource, and has not been rigorously evaluated for target organ predictivity. Tox21BodyMap should therefore not be used as the basis for chemical hazard characterization or risk assessment, and is solely intended for exploratory purposes and hypothesis generation.

Go to Tox21BodyMap

The Threshold of Toxicological Concern (TTC) approach has been developed to qualitatively assess the risk of low-level substances in the diet. It can be used for an initial assessment of a substance to determine whether a comprehensive risk assessment is required. It is an important science-based approach for prioritising assessment of those chemicals with low-level exposures that require more data over those that can be presumed to present no appreciable human health risk.

Scientists have developed various decision trees based on the concept of TTC. The most well-known decision trees are Cramer and Kroes. The decision tree is based on the structure and toxicity data of chemical substances to set up multiple questions, which can ultimately determine the level of concern about the toxicity of the substance. Software companies have integrated various methods. In addition to Cramer and Kroes decision trees, it also includes carcinogenicity prediction, skin irritation prediction... and other prediction methods.

EFSA: Threshold of toxicological concern
WHO: Threshold of Toxicological Concern (TTC)
Download Toxtree Download reference