DescriptionThe serotonin type-6 receptor (5-HT6) is a drug target for psychotic diseases, especially cognitive disorders. The traditional method to design novel 5-HT6 binding agents (e.g. antagonists) is to experimentally screen a large chemical dataset that is randomly selected from a drug-like chemical library. This process is normally very costly and has a low success rate. Computer Aided Drug Discovery (CADD) uses computational models to virtually screen a chemical library and select promising candidates for experimental testing. Using CADD, the resources could be saved and the success rate could be increased by excluding unsuitable compounds. Quantitative Structure-Activity Relationship (QSAR) is the most frequently used method for developing various predictive models within the drug discovery process. In this work, a 5-HT6 dataset of 488 unique compounds was compiled. Among them, 225 were experimentally identified as 5-HT6 antagonists and the remaining were diverse anti-cancer compounds, which were considered to be unable to bind to 5-HT6. I applied various QSAR modeling approaches to develop several computational binary 5-HT6 models. The resulting models were validated by a five-fold cross-validation approach and the resulting predictivity, which was measured using Correct Classification Rate (CCR), was 96%. The resulting models to predict an external data set and the predictivity (CCR=88%) was similar to the cross validation. Thus, the models developed in this study could be used to detect novel 5-HT6 ligands in the future drug discovery process.