MotifVisualizer: An Interdisciplinary GUI for Geometrical Motif Retrieval in Proteins

摘要

In previous works, we presented Cross Motif Search (CMS), an algorithm designed to explore new techniques in the field of protein motif retrieval and identification. The novelty of the CMS approach is to look for geometrical similarities in the secondary structures of proteins, instead of homologous topology. Put in other words, while the connections among different secondary structures are still considered, CMS puts the emphasis in the overall 3D shape of candidate motifs and identifies them using computer vision techniques, such as a modified Generalized Hough Transform (GHT). This approach, even if slower than classic approaches such as PROMOTIF or ProSMoS, and less resistant to deformations, has two distinct advantages: first of all is able to conduct a precise detection, that is, is able to exactly identify the positions of each candidate, most importantly, is also able to identify similar geometrical structure even in unfamiliar proteins (i.e. proteins belonging to different families in the SCOP database) and similarities which may elude topological algorithms. Unfortunately, while we were able to accumulate a huge database of results, we are still working on classifying the matches according to their significance. Indeed, one aspect which is often overlooked by computer scientists working in this field is the cross-cooperation with biologists, which usually do not speak the same common language of engineers, and vice versa. In this paper, we describe the MotifVisualizer application and the algorithm design of CMS. MotifVisualizer has been conceived to be an easy-to-use graphical user interface to the CMS algorithm, and through it we hope to close the gap between engineers and biologists, which may help us to showcase the significance of the CMS matches.