Summary of Our
Research
We work in the area of computational analysis of DNA and protein, sequence and structure data to understand their function. We have been developing computational tools and web servers that help to provide value added data resources and useful results of analyses. Our research publications cover broadly the DNA sequence data analysis, protein sequence and structural data analyses, protein modeling, structure prediction and docking studies. These studies broadly fall in the area of "Computational Biology and Bioinformatics" (see list of publications below for all the references cited here). The recent technologies in genomic sequencing, high throughput structure determination and microarray techniques have been generating wealth of biological data. The key issues in building reliable knowledge-bases (expert systems) are efficient data mining and finding the right differentiators that are unambiguous, quantifiable and sufficiently robust to deal with the heterogeneous datasets common to biological systems. We consider the field of bioinformatics is very challenging since it involves a combination of tasks such as thorough understanding of biological processes, expert-level data mining and interpreting, designing the algorithm to extract information, developing predictive methods and making databases and analysis tools accessible on the web servers. In the past we developed several useful computational methods using both sequence and structure data analyses and have published several well cited research findings summarized below. A. Protein Structure Data Analyses: Protein-Protein Interactions: We have been working to understand protein-protein interaction, analyzed several known protein-protean complex structures and computed various parameters that help in predicting the interaction sites between the two protein structures. We have done docking calculations, computed pair potentials, desolvation energy, conservation index and several other parameters to develop computational method to predict interacting surface regions of protein structures (Reddy et al. 2005; Duan et al., 2005; Duan et al., 2006). Common Substructures: Our work on analysis of common substructures in proteins given a method that utilizes structure based sequence alignments in the protein data bank (PDB) and protein sequence database (UNIPROT, NR). The method identifies conserved key amino acid positions (CKAAPs) for each commonly occurring substructures and representative protein structures in the PDB. CKAAPs are proposed to be useful in fold recognition and protein engineering. (Reddy et al.,2002; Reddy et al., 2001). Using this method we have developed a CKAAPs resource database (Li et al., 2001, Li et al.,2002). Secondary Structural Packing: Geometry of packing of secondary structural elements in proteins has been studied on helix-helix, helix-sheet, and sheet-sheet packing. The results were shown to be useful to improve comparative protein modeling procedures. The work helps in adjusting the rigid secondary structural elements, helices and sheets, in the core of the protein models (Reddy, 2002; Reddy & Blundell, 1993; Reddy et al., 1999; Nagarajaram et. al, 1999). Protein Stability: We have evaluated the structural environment of amino acids in known protein structures. About 360 substituted single residue mutants were collected from the literature for which parent protein structure and altered in vitro (thermal, solvent or pH induced) stability information was available. The structural environment of each of the parent residue is characterized and a method is developed to suggest substitution mutations to engineer in vitro stability of a given protein sequence for which previously determined structure is available (Reddy et al., 1998). Modeling Studies: We have done comparative modeling of structure of rabbit M4-Lactate Dehydrogenase, based on the available structure of dogfish LDH-M4 (Rajenderkumar et al., 1994). We participated in the CASP3 test and succeeded with best models (Burk et al., 1999). We also submitted CASP3 targets for secondary structure prediction using the method we developed (Tiwari & Reddy, 1999). We developed a modeling procedure for DNA-protein interaction using homology based prediction and chemical and stereo-chemical rules from the knowledge of known X-ray/NMR structures of DNA-protein complexes. A general method of approach is standardized to model sequence dependent DNA-binding. Using this approach we modeled a DNA-protein complex in the region of -35 hexamer of promoter interaction with 4.2 helix-turn-helix domain of sigma-70 subunit of E. coli polymerase and its mutants V576G, V576T (Reddy et al., 1997). Peptide Modeling
and Drug Design:
We have done
comparative modeling of small peptides that were identified to mimic scatter
factor and FGF. Later small molecular compounds were screened from available
chemical directory and designed analogous chemical compounds to the peptide
model structures. Four of the compounds were tested experimentally and found
to be potential drug compounds by a B. Protein
Sequence Data Analyses and Protein Stability: Proteins are known to degrade rapidly when conformations are altered due to abnormality in the sequences. Normal cellular proteins also display a wide range of half-life - turnover rates of individual proteins can differ as much as 1000-fold. Sequence specific properties, global features and the location of a protein in the cell are found to be important in deciding the intracellular stability of a protein. In order to identify sequence dependent properties we have analyzed the stable and less stable protein sequences and observed that the di-peptide composition in stable proteins is significantly different from the less stable proteins. This observation was used to develop a computational method to predict protein stability from its amino acid sequence information (Guruprasad et al., 1990). We further analyzed the structural location of these di-peptides in non-homologous protein structures and found that the general distribution of sensitive (stable and less stable) di-peptides is high in the regions that are on the surface and have more hydrogen bonding interactions with near neighbor residues. These di-peptides are also usually present closer to the molecular surface and significantly more occurrence in turns and random coil structures (Reddy, 1996). Based on these structural observations we developed a method that gives theoretical suggestions for substitution mutations to alter the intracellular stability of a given protein. Intelligenetics USA (PHY_CHEM) has integrated our algorithm into the sequence analysis software, PC-GENE package. The method further refined and developed as bioinformatics tool to suggest substitution mutations to increase intracellular stability of proteins (MEICPS: Reddy et al., 1999; Sheri et al., 2005). C. DNA Sequence Data Analyses: Prokaryotic DNA Sequence Data Analyses: We have carried out analysis of DNA sequence data to understand synonymous codon usage (Kolaskar & Reddy, 1985a). We developed a method to locate protein coding sequences in DNA of prokaryotic systems (Kolaskar & Reddy,1985b). This algorithm has been integrated in the PC-GENE DNA sequence analysis software package by Intelligenitics USA (COD_PROC). We have further studied the contextual constraints on codon pair usage and also synonymous codon usage in the DNA sequence (Kolaskar & Reddy, 1886; Kolaskar et al., 1995). Eukaryotic DNA Sequence Data Analyses: We have also analyzed the Mice and Human gene sequences. We developed a statistical analytical method to predict splice-sites in these DNA sequences. The method has given best prediction results with much less false positives compared to the other methods available during the time (Reddy et al., 1991; Reddy & Pandit, 1995). --------oOo-------- |
List of Publications:
Book Chapters:
Published
Abstracts:
|