Current projects:

Comparative genomics and phylogenetics
Theoretical and practical aspects of evolutionary relationships between genes and species are an important topic of our research. For instance, we have recently developed URec software for comparing evolution of genes and species and infering species phylogenies.
Evolution of gene families
We work on different models of evolution of paralogous gene families in genomes. We have recently proposed a discrete probabilistic model that is based on four evolutionary events: gene duplication, loss, accumulated change and innovation. We have focused on the size distribution of gene families, and shown that their sizes follow a logarithmic distribution.
Evolution of protein-protein interactions and network alignment
We are interested in mathematical models of network evolution and how they can be used to compare PPI networks across species. We have recently designed a new evolutionary based approach to network alignment. The method uses Bayesian modeling to reconstruct a hypothetical ancestral PPI network and identifies highly conserved modules in the networks of present-day species.
Regulatory sequence analysis
We are interested in improving methods for transcription factor binding site motif analyses. We developed MMF - Meta Motif Finder for finding consensus solution to motif finding problem by clustering results of different de novo methods. We also developed a new method for finding instances of a given motif taking into account negative information. Currently, we work on a framework for finding loosely conserved (Billboard type) cis-regulatory modules in non-coding DNA regions of homolgues genes.
Regulatory network reconstruction
We work on different computational approaches to reconstruction of regulatory gene networks from experimental data. We focus on probabilistic methods such as Bayesian networks (static and dynamic) and Stochastic Logical Networks. For Bayesian Network inference, we proposed a new efficient algorithm. We also developed a rule-based method for aggregating the data on co-expression with sequence motifs.
Contextual sequence alignment
The model of contextual alignment is an extension of the classical alignment, in which the cost of an amino acid substitution depends on the surrounding residues. Efficient algorithms for optimal contextual alignment were developed, including CTX-BLAST that incorporates contextual alignment model into the popular protein BLAST program.
Mining DNA transposons
We have recently developed a specialized tool, called TIRFinder for mining DNA Class II transposons in sequenced genomes, on the basis of the structural characteristics of a particular transposable element family. Using this tool we characterized PIF/Harbinger-like elements in the genome of Medicago Truncatula.
Analysis of mass spectrometry data
We work on computational methods supporting medical diagnosis by peptide mass spectrometry. Our tools enable for efficient processing of mass spectra and discriminant analysis of LC-MS samples. New consensus biomarker approach is proposed for aggregating different feature selection methods.

We pariticipate in the following grants: