Biennial Report 2011–2012

At present computational methods are playing an increasingly
important role in biological research. Breakthroughs in
technologies have resulted in a flood of various types of biological
data such as genome sequences for different organisms,
data on gene expression, protein-protein interactions,
etc. Computational biology and bioinformatics are helping to
make sense of all this vast biological data by providing tools
for performing large-scale studies. In addition, computational
biologists are utilizing available experimental data to improve
various analytical and predictive methods that could help address
specific biological problems.
Research carried out in our department covers a broad range of
topics that together can be described as Computational Studies
of Protein Structure, Function and Evolution. There are two main
research directions:
Development of methods for the detection of protein homology
from sequence data, comparative modeling, analysis
and evaluation of protein three-dimensional structure.
Application of computational methods for discovering
general patterns in biological data, structural/functional characterization
of proteins and their complexes; design of novel
proteins and mutants with desired properties. We address a variety
of challenging biological problems, yet our main focus is
on proteins and protein complexes involved in DNA replication,
repair and recombination.
Development of
computational methods
At present, we are in the final stages of the development of a
new competitive homology detection method.
The evaluation of protein structure is particularly important in
computational protein modeling. Scoring models against the
native structure is at the heart of development and benchmarking
of protein structure prediction and refinement methods. It
may seem that one-to-one correspondence between computational
models and the native (reference) structure should make
such evaluation trivial. Yet, contrary to this view, it is an open
problem, because many aspects of the reference-based model
evaluation still lack desired robustness. We have been actively
researching how to improve the reference-based model evaluation.
Our attempts resulted in a new highly effective score,
which is described in more detail below.
CAD-score: evaluation of protein structural
models based on contact area difference
CAD-score (Contact Area Difference score) is a new evaluation
function quantifying differences between physical contacts
in a model and the reference structure. It uses the concept
of residue-residue contact area difference (CAD) introduced
by Abagyan & Totrov (J. Mol. Biol. 1997; 268:678–
685). Contact areas, the underlying basis of the score, are derived
using the Voronoi diagram of spheres that correspond to
heavy atoms of van der Waals radii (Figure 1). The Voronoi diagram
of spheres is constructed by a new algorithm that is especially
suited for processing macromolecular structures.
During the report period our major efforts in methods development
were devoted to protein homology detection and evaluation
of protein structure.
The concept of homology (common evolutionary origin) is at
the heart of most studies dealing with protein sequence, structure
and function. In the absence of three dimensional (3D)
protein structures the homology detection has to rely on sequence
data. Currently, the most sensitive homology inference
methods are based on comparison of multiple sequence alignments
represented as sequence profiles. However, these profiles
can be constructed, compared and scored in many different
ways. On the one hand this complicates the development of
profile-based homology detection methods; on the other hand
this means a lot of space for improvement. During the two
years covered by this report we have been actively exploring different
paths to improve the profile-based homology detection.
Figure 1. Voronoi diagram of spheres of a residue pair
The algorithm resolves residue–residue contacts at the level of
atoms, making it possible to consider contacts not only between
entire residues but also between subsets of residue atoms
(main chain, side chain) (Figure 2).
Vilnius University Institute of Biotechnology Biennial report for 2011–2012 65