Improving pKa Prediction Accuracy for PROTACs

This whitepaper evaluates the predictive performance of the ACD/pKa Classic algorithm specifically for PROTACs (Proteolysis Targeting Chimeras) and their precursors. The study highlights the challenges in predicting properties for compounds that fall outside the traditional Lipinski 'Rule-of-Five' criteria, particularly focusing on the pKa values of these complex molecules. A dataset of 253 PROTACs was analyzed, with 491 experimental pKa values collected from various sources. The results indicate that the ACD/pKa Classic algorithm demonstrates strong performance in predicting pKa values, especially with the incorporation of PROTAC data into newer software versions. The evaluation also emphasizes the importance of continuous data collection and curation to enhance prediction accuracy. The findings suggest that the algorithm can reliably predict pKa values for new therapeutic modalities, thereby supporting the development of drugs that do not conform to traditional chemical space guidelines.