Multi-view Spectral Clustering and its Chemical Application: Clustering TCAMS and ZINC Database Compounds

publication · 6 years ago
by Adeshola A. Adefioye, Xinhai Liu, Bart De Moor (Katholieke Universiteit Leuven)
Here an improved spectral clustering algorithm is proposed for chemical compound clustering. Clustering serves an essential role in chemoinformatics. It is an unsupervised method that allows researchers to learn information about compound relationships with other compounds. With the clustering results, it is possible to hypothesize about a compounds biological, chemical and physical property when compared to another compound. Using the kernel and tensor based spectral methods, as done in this paper, provides chemically appropriate and statistically significant results when attempting to cluster compounds from the GSK-Chembl Malaria data set as well as the Zinc database. Methods such as spectral clustering, proves to be essential to aiding researchers gain greater insights when searching for similarities between and amongst a compound set. Spectral clustering algorithms based on the tensor method give robust results on the mid-size compound sets used here.
Visit publication