The Mapper Algorithm for the Qualitative Analysis of Drug Molecules
MetadataShow full item record
The world of computational drug design has seen tremendous progress in the last 40 years. Most of these methods rely on quantitative techniques through quantitative structure activity relationships. This approach works well though it fails to reflect the qualitative nature of drug design. Here in we investigate the Mapper algorithm for analysis of drug molecules, a recent development in topological data analysis that offers a qualitative view of a data set. The Mapper algorithm is first used to look at drugs with respect to their toxicities via equal length intervals and intervals determined by CLINK (Complete LINKage) clustering. The algorithm is then used to look at the similarity to drug structures compared to some lead compound. The compounds are analyzed with regards to their dissimilarity, or similarity, from this lead structure. The algorithm is able to form clusters of drugs which share common moieties especially in the case of morphine derivatives. Overall the algorithm shows promising results which merit further research.