Large scale ontology based latent semantic indexing (OLSI) of scientific publications to create toxicology prediction systems

Presentation held at the zeroPM workshop on September 19, 2024. We demonstrated the utility of chemical ontologies for classifying compounds. Here, we created a organic chemistry ontology and classifying about 1.7 million poly and perfluorinated compounds (PFAS) found in scientific publications and patents (https://pubchem.ncbi.nlm.nih.gov/classification/#hid=134). These indexed publications were used to retrieve PFAS - Toxicity cooccurrences. The statistics of these coocurrences were than used to create a prediction system for PFAS compounds to aid risk management when dealing with these "forever" compounds.