Although my research program has several foci, the common theme among them is the use of mathematics and computing in solving current problems in biology and medicine. My research program generally targets problems that arise from the ‘sequence-structure-function’ paradigm in molecular biology. We usually approach the problems from the structural angle, as structure plays a more direct role in effecting function, at least for proteins and their variant biomolecules in the cell. Thus we typically work with coordinates rather than with sequences, and our main dataset are the nearly 76,000 experimentally determined structures in the Protein Data Bank to date, although sometimes we do work with predicted (modeled) structures. We use a multidisciplinary approach in solving the problems that we do address, and we always try to create our own algorithms and build our own tools, with Fortran and Perl in a UNIX command-line environment as our main workhorses. A sampling of the types of questions we address are enumerated below:
- given a protein’s 3D structure, how can one predict the location of the binding site, if any, of a specified ligand on the protein surface?
- how can one predict the location of interaction sites of a given protein with other proteins given its 3D structure?
- how can we synthesize the information from the two exercises above to predict the protein’s biological function?
- is there a general and automatable way to project the surface of a protein onto a plane that will allow for meaningful comparison of proteins surfaces?
- how can we quantitatively measure the difference between the 3D structures of any two given proteins that may or may not be of the same size?
- what are the ‘rules of association’, if any, among the secondary structures (α helices, β strands/sheets, and loops) of a protein in assembling its 3D structure?
- given the 3D structure of a protein, how can one predict the location of epitopes, if any, on its surface?
- how can protein structures be more effectively represented using other coordinate systems besides the most commonly used Cartesian coordinate system?
- is there a common motif for specific post-translational modification sites, e.g., lipidation, glycosylation, phosphorylation sites, on protein surfaces?
- what in general are the differences in function of the amino acids in the outer layer of a protein versus those in the inner core?
- would a multi-origin coordinate system, such as the bispherical system, be useful in systematically isolating the domains in a multi-domain protein?
The above is a partial and evolving list, and from time to time we add questions that may have arisen out of working on a current problem, or out of sudden curiosity, and also take ones out that have already been addressed or have hit a dead end.
Some of the algorithms we have developed have been implemented in webservers in order to allow the community to use and/or test our algorithms, as follows:
• In The Protein DCRR Webserver (with Vrunda Sheth), URL: http:// tortellini. bioinformatics. rit. edu/ vns4483/ dcrr.php, the user enters the PDB ID of the protein to be transformed in DCRR (‘double centroid reduced representation’), and the webserver outputs the DCRR file (similar to a PDB file but coordinates are those of backbone and sidechain centroids instead of atoms); the user can also browse for and input a PDB file form his/her own directory (e.g., a predicted structure). In addition, the webserver also outputs a MatLab graphics file that displays the protein structure in DCRR (i.e., centroids instead of atoms) using MatLab.
• In The Secant Plane and Tangent Sphere Webserver (with Srujana Cheguri), URL: http:// tortellini. bioinformatics. rit. edu/ sxc6274/ thesis1.php, the user enters the PDB coordinate file of a protein receptor, and the ligand coordinate file or the coordinates of one or more amino acids in the ligand binding site, and the webserver outputs the Secant Plane and Tangent Sphere indices (‘SPi’ and ‘TSi’, respectively), which are measures of the ligand binding site burial in terms of the two complementary methods.
• In The Protein Cylindrical Coordinates Webserver (with Srujana Cheguri), URL: http:// tortellini. bioinformatics. rit. edu/ sxc6274/ thesis2.php, the user enters the PDB coordinate file of a rod-shaped protein and the coordinates of the two points at the extremities of the protein and the webserver outputs the protein structure in cylindrical coordinates with the z-axis as the main axis and one of the extremity points at the origin.
We have two additional webservers in the works, namely, The Protein Structure Spherical Metric Webserver (with James DeFelice), tentative URL: http:// tortellini. bioinformatics. rit. edu/ jpd3501 /sphmetric.php, and the The Cancer Biotherapy Database/Webserver (with Preety Priya), tentative URL: http:// tortellini. bioinformatics. rit. edu/ pxp4438 /cancerbiother. php, which we hope will be of help to the community just as the three previously mentioned webservers are. We plan to build publicly accessible webservers that implement our other algorithms at the time of publication of the relevant paper.
The following abstracts and posters which members of my research group and I have presented in various scientific conferences since January 2009 highlight specific research problems that we are currently trying to tackle. Some of these abstracts currently have manuscripts in preparation. To see the abstracts and posters, please click here. |
