Carcinogenesis Predictions Using Inductive Logic Programming

Abstract

Obtaining accurate structural alerts for the causes of chemical cancersis a problem of great scientific and humanitarian value. This chapterbuilds on our earlier research that demonstrated the use of InductiveLogic Programming (ILP) for predictions for the related problem ofmutagenic activity amongst nitroaromatic molecules. Here we areconcerned with predicting carcinogenic activity in rodent bioassaysusing data from the U.S. National Toxicology Program conducted by theNational Institute of Environmental Health Sciences. The 330 chemicalsused here are significantly more diverse than the mutagenesis study, andform the basis for obtaining Structure-Activity Relationships (SARs)relating molecular structure to cancerous activity in rodents. Wedescribe the use of the ILP system Progol to obtain SARs from this data.The rules obtained from Progol are comparable in accuracy to those fromexpert chemists, and more accurate than most state-of-the-art toxicityprediction methods. The rules can also be interpreted to give cluesabout the biological and chemical mechanisms of cancerogenesis, and makeuse of those learned by Progol for mutagenesis. Finally, we presentdetails of, and predictions for, an ongoing international blind trialaimed specifically at comparing prediction methods.