Computational Methods for Paleogenomics and Comparative Genomics

Genome evolution

Our main research theme is to develop computational methods to analyse extant genomes in a phylogenetic context, to understand the process that shape genomes during evolution. For this, we focus on several problems, such as reconstructing the gene order of ancestral genomes, computing genome rearrangement scenarios, computing evolutionary scenarios for gene families, ...

One of our favorite research problems is the reconstruction of ancestral genome structures (genome maps based on synteny blocks, or gene orders). Our approach follows the comparative paradigm, that assumes that conserved features of related extant genomes indicate potential ancestral genome features. We work within two methodological frameworks for this problem: a local approach, that considers a single ancestral genome within a given species phylogeny [PLoS Comput Biol 2008,ANGES], and a global (aka small parsimony approach), that considers all ancestral genomes of a species phylogeny at once [ISBRA 2016.

Lately, we aim to extend these approaches in order we can incorporate in the computation models genefamily events such as gene duplication, loss or transfert. In this approach, we developed a probablistic approach within the DeCo algorithmic framework [DeClone]. This line of work also motivates motivated a series of papers on the correction of gene trees [Bioinformatics 2014] and on the reconciliation between gene trees and species trees [ecceTERA].

The recent breakthroughs in ancient DNA (aDNA) sequencing naturally lead us to investigate the problem of applying the principles of the methods we developed to reconstruct ancestral genomes maps to the assembly of aDNA sequencing data. Our first results, on scaffolding of ancient Yersinia pestis genomes illustrates the relevance of this approach. [FPSAC].

We are currently focusing a lot on the analysis of a fascinating, large-scale, data set composed of toughly twenty Anopheles mosquito genomes [Science 2015]. This in turn raises interesting questions on how to handle fragmented genome assemblies in genome rearrangement studies [BMC Genomics 2015].

Last, in a starting collaboration with Leonid Chindelevitch and Will Hsiao, funded by Genome Canada, we are interested in the evolution of human pathogens, either within short-time local outbreaks, or within a large historical context. This work ties up nicely with our current work on the analysis of ancient human pathogens using ancient DNA data.

Big data flow cytometry bioinformatics

This topic is new in our lab. It stems from a starting collaboration with Ryan Brinkman (BC Cancer Agency) and Sara Mostafavi (UBC), funded by Genome Canada. Stay tuned, we will likely have results to discuss here soon.

RNA secondary structures comparison.

We are mostly interested in developing accurate and efficient algorithm to compare pairs of RNA secondary structures. Among others, we have been involved in the development of the BRASERO benchmark, the analysis of the BRALIBASE benchmark [Briefings Bioinformatics 2016], and in the design of efficient dynamic programming algorithms for comparing RNA secondary structures [AlCoB 2016].

Our most recent work, in collaboration with Bordeaux University, showed that the chaining approach, widely used in sequence comparison, can be extended very to RNA secondary structures to offer an efficient and accurate method to find structure-sequence homologs in a database of RNA sequences [JDA 2012; JCB 2015].