The Development Of Two Computational Methods And Their Web Servers To Support Drug Safety Evaluation

Drug safety evaluation is one of the core problems of drug preclinical studies and it has caused the attention of drug regulators and academia. In consideration of the high cost of experiment and animal ethics, alternatives to animal testing such as computational prediction method have been paid much attention. This study has established two prediction methods of drug safety analysis and platforms based on the toxicogenomics and DITRPs (drug-induced toxicity related proteins) respectively. The main work and results are described as follows:(1) Here we established a safety prediction method for evaluating the potential liver toxicity based on genome-wide transcriptomics data and developed a computational platform, namely LTMap.Applying the Kolmogorov-Smirnov statistics developed by Lamb et al., the rank ordered mode matching strategy would be used to analysis signatures induced by the query compound with signature databases derived from Open TG-GATEs. Combined withcase studies of the external validation, we discussed the ability of the computational approach in our studyto revealhepatotoxicitymechnism between chemicals and discover unexpectedtoxicological responses. Furthermore, a user-friendly web interface is provided by LTMap to browse and search toxicogenomics data.(2) We developed a safety assessment method based on the target-ADR relations to evaluate the safety liabilities of drug candidates and built an online analysis system TADRR. Firstly, applying the similar ensemble approach algorithm developed by Keiser et al., we established target prediction system with the known DITRPs and there corresponding ligands on a large scale. Then we connected the targets and ADRs based on the drugs they are associated using the method called 'guilt-by-association'.According to the drug-tatget-adverse relations, this method could be used to predict potential safety liabilities. TADRR analysis platform was developed basedon the above method. Simply submita compound with its 2D chemical structureto TADRR, potential safety liabilities would be suggested according to TADRR.