Projects / Programmes
Algorithms development for protein binding sites prediction
| Code |
Science |
Field |
Subfield |
| 1.07.00 |
Natural sciences and mathematics |
Computer intensive methods and applications |
|
| Code |
Science |
Field |
| P000 |
Natural sciences and mathematics |
|
| Code |
Science |
Field |
| 1.07 |
Natural Sciences |
Other natural sciences |
Protein binding sites, binding sites detection, structural alignment, bioinformatics, drug development, graph theory
Researchers (1)
| no. |
Code |
Name and surname |
Research area |
Role |
Period |
No. of publicationsNo. of publications |
| 1. |
25435 |
PhD Janez Konc |
Computer intensive methods and applications |
Head |
2010 - 2012 |
236 |
Organisations (1)
| no. |
Code |
Research organisation |
City |
Registration number |
No. of publicationsNo. of publications |
| 1. |
0104 |
National Institute of Chemistry |
Ljubljana |
5051592000 |
21,023 |
Abstract
The development of molecular modeling methods for studying protein-protein interactions is becoming increasingly important, since interactions between proteins play an important role in most biological systems. Prediction of protein binding sites is particularly important for the developement of new drugs targeting protein-protein interactions. Binding sites are functionally important regions of protein surfaces, through which proteins interact with their partners. The specificity of protein-protein binding is strongly related to the conservation of protein binding sites in related proteins. Despite rapidly increasing number of protein structures deposited in the protein data bank, there is still a relatively small number of known structures of protein complexes. Knowing the location of protein binding sites is important for understanding protein-protein interactions and can be used to improve modeling structures of protein complexes. Pharmaceutical industry is currently using about 150 protein targets for drug development, although an estimated of 6000 targets with therapeutic potential exist. Since binding sites are closely related to protein function, an approach that predicts binding sites on proteins could be used to discover new therapetucally interesting protein targets.
The goal of the proposed research is the development of new algorithms for prediction of protein-protein binding sites, which will be based on the fact that protein surface structures are conserved in binding sites regions. These algorithms will search for local similarities in physico-chemical properties in different protein surface structures independently of sequence or fold. The proteins will be modeled as protein graphs, i.e., rigid 3D objects, consisting of vertices and edges. The newly developed algorithms will be used for prediction of protein binding sites on known structures of proteins from the RCSB protein bank; particularly, we will focus on pharmaceutically interesting proteins, which could become targets for the next generation of pharmaceuticals, e.g., inhibitors of protein-protein interactions. Predicted binding sites will also be used to enhance modeling of structures of protein complexes using a protein-protein docking approach.
Our developed algorithms for prediction of protein binding sites will be open to public in the form of our freely accessible ProBiS (Protein Binding Sites) web-tool, available at http://probis.cmm.ki.si, which will contribute to wide usage of our methods.
Significance for science
Algorithms for prediction of binding sites on proteins enable new insights into actions of proteins and are fundamental to our understanding of how ligands (small organic or inorganic, proteins, nucleic acids) bind to proteins. Our newly developed methods for binding sites detection and structural comparison of binding sites can predict biochemical functions for proteins whose activity is unknown. Due to increasing number of solved, but uncharacterized, protein structures in the PDB that come from structural genomics projects, novel approaches to functional annotation of proteins are needed. Our methods are part of the solution to this challenge.
We developed a web server ProBiS (http://probis.cmm.ki.si) for the detection of structurally similar protein binding sites on proteins without any reference or prior knowledge of these sites, which was not possible with existing tools. The web server is based on the developed ProBiS algorithm for local structural alignment of protein structures, which uses our efficient maximum clique algorithm on protein graphs. ProBiS web server is freely accessible on the world wide web, and has been heavily visited since the publication of the corresponding papers: monthly, we count around 500 visits, mostly from the USA (source: Google Analytics). ProBiS can be applied to structural and systems biology for the functional annotation of uncharacterized protein structures. It uses little computer time for the local structural alignment of complete protein structures and therefore extends the applicability of such comparisons to large structural databases such as the RCSB Protein Data Bank. In addition, pharmaceutical research, particularly the structure-based drug design process, benefits from the unique ability of our local structural alignment method to align binding sites in proteins with dissimilar folds. This enables the detection of similar and dissimilar features of such aligned binding sites and development of specific new compounds targeted at particular organisms.
Upon an invitation of the RCSB Protein Data Bank authors, the ProBiS server has been listed among the protein structure analysis tools found at their website, and has also been included in the CCL (Computational Chemistry List), where it was listed among the tools for the computer analysis of protein structures. Here, the most successful programs for the classification of newly determined proteins, especially those from structural genomics projects, are collected on a common site.
The links to the aforementioned websites are:
http://www.rcsb.org/pdb/static.do?p=general_information/web_links/structure_classification.html
http://www.ccl.net/chemistry/links/software/index.shtml
Significance for the country
The goal of our project was the development of algorithms for detection of protein binding sites. We developed a new algorithm for local structural comparison of protein structures, and on it, built a new web server, ProBiS. We applied ProBiS to pharmaceutically interesting proteins from structural genomics.
Our algorithms are of interest to science as well as to pharmaceutical industry. Accordingly, we made the ProBiS web server graphically appealing and user friendly and open to public. The results of our research has been published in international scientific journals. Our methods can compare and sometimes outperform other state-of-the-art methods and are freely available, to increase our country's international recognition as well as sustain and enhance its national identity.
We have collaborated on a common project entitled Modeling of biopharmaceutical molecules, with Lek d.d., a slovenian pharmaceutical company. In this work, we successfully used ProBiS web server as a tool for predicting and modeling the biological activity of certain biopharmaceutical molecules. Our expertise in fundamental science of binding sites detection was here used to solve a very applicative task, and thus we contributed to a fast transfer of knowledge between our Laboratory for molecular modeling, and the slovenian pharmaceutical industry.
Most important scientific results
Annual report
2010,
2011,
final report,
complete report on dLib.si
Most important socioeconomically and culturally relevant results
Annual report
2010,
2011,
final report,
complete report on dLib.si
