OUR DATA

Regular Updates and Advanced Data Mining

Our databases are systematically updated, drawing from a comprehensive array of both open and private sources.

The chemical space is represented with pharmaceuticals, industrial chemicals, pesticides, biocides, flavoring agents, food additives, cosmetic ingredients, natural products and other diverse chemicals used in different R&D. 

We leverage a customized Large Language Model to meticulously extract pertinent information from over 35 million scientific papers, significantly enhancing the quantity and diversity of our databases.

 

Data quality

The quality of data plays a pivotal role in the development of a robust and reliable QSAR model. All our datasets are manually reviewed and normalized by experts, adhering to the Klimisch criteria. The Klimisch criteria is founded on a scoring system designed to evaluate the reliability of data. This rating system was initially formulated for toxicological and ecotoxicological studies. Subsequently, it has been expanded to encompass physicochemical studies and is currently recognized and accepted by numerous regulatory authorities and organizations.

Following manual expert review, the datasets, representing endpoints or properties outlined in regulatory guidelines and aligning with standardized test procedures, receive a score of 1, placing them within the 'reliable without restriction' category. Our experts retain only data that aligns or is comparable to guideline studies, preferably conducted according to GLP, ensuring that test procedures adhere to national standard methods and generally accepted scientific standards, all described in detail. The datasets representing endpoints or properties obtained in non-GLP studies, but conducted comparable to guideline studies with acceptable restrictions and/or utilizing test procedures in accordance with national standard methods, also with acceptable restrictions, or well-documented studies adhering to generally accepted scientific principles and deemed acceptable for assessment, or utilizing accepted calculation methods, receive a score of 2 and are categorized as 'Reliable with restriction' data. 

​

​