|
|
|
|
|
|
Current scientific interests and former research projects
- Scientific visualization and molecular modeling (1996 - present)
Besides my professional interests, molecular modeling and visualization is also my hobby.
In my spare time I have been developing free, open source scientific visualization programs: Biodesigner and iMol.
- Structural DNA nanotechnology (2008)
Structural DNA nanotechnology (SDN) is a very cool branch of knowledge that allows for manufacturing of molecular-scale
objects with atomistic precision. The array of possible applications is vast, ranging from atomistic-size
devices to molecular scaffolds to DNA-based computers.
In 2008 I worked for Nanorex who created one of the most powerful DNA modeling tools: molecular CAD program NanoEngineer-1. The program is free and open source, check it out!
- Analysis of conformational changes within protein families (2006 - 2007)
Three-dimensional structure of proteins often changes during the course of evolution
to facilitate novel functions.
The structure of protein can also change while the molecule takes part in various metabolic processes.
Are these structural changes related? I tried to answer this question by developing methods
for analysis and comparison of the conformational changes in proteins. My 3D structure interpolation method
is now available as a part of FATCAT Flexible Alignment Server
at Burnham Institute for Medical Research.
- Converting reduced representations of protein structures to full-atom models (2005)
In order to efficiently sample conformational space of proteins, prediction algorithms often employ reduced protein representations.
While the reduced models are necessary to efficiently search conformational space, fine-detailed all-atom models are often essential
for subsequent structural studies such as protein function analysis, virtual ligand screening, prediction of protein-protein
interactions, and assessment of the structure quality. Therefore, the reduced models need to be accurately translated into detailed
atomistic models. I developed a program called PULCHRA (it stands for
PowerfUL CHain Restoration Algorithm) to automate and simplify this task.
- Modeling of vitamin D receptor (2000 - 2008)
I designed a very accurate model of vitamin D receptor long before the actual experimental structure was known.
I assisted at subsequent theoretical analyses of biological activity of the vitamin D derivatives.
Several little tools were created as an offspring of the project, including CCOMP - receptor/ligand complexes comparison tool.
- Computational evaluation of mimetic petides (2004)
Proteins found on a surface of tumors such as neuroblastoma or melanoma can induce weak response of
immunologic system. Similar response can be induced using short peptides similar to certain
parts of these proteins (the mimetic peptides). To speed up experimental research in this area,
I developed a computational framework for testing binding affinity of the mimetic peptides
to immunoglobulin protein.
- Design and evaluation of statistical potentials for protein structure prediction (1996 - 2002)
Protein structure modeling methods usually incorporate information derived from known protein structures.
This information can be represented in a form of statistical (or knowledge-based) potentials.
During that period I was particularly interested in homology-enhanced statistical potentials.
These potentials became part of SICHO/MONSSTER/CABS family of protein structure prediction programs, developed
over the years by prof. Andrzej Kolinski at Warsaw University.
- Comparative modeling of protein structure using distant homologs (1998 - 2002)
Range of applicability of protein modeling methods depends on a level of
similarity of the modeled sequence to the structure identified as a modeling template.
Typically, these methods tend to fail if the sequence identity level between the target
and the template falls below 25%. By using homology-enhanced statistical potentials
and lattice Monte Carlo simulation engine, this practical limit of protein structure prediction
methods can be moved towards very low sequence similarity levels.
- Protein structure prediction (1998 - 2002)
I participated in blind protein structure prediction experiments:
CASP3,
CASP4, and
CASP5.
Our group's performance was excellent and we delivered some of the best predicted protein models.
|
|
