Grant: Discovery Grants Program - Individual

We study structural genome evolution in terms of gene order, concentrating on the flowering plants. One focus ison branching process models for the expansion and contraction of single gene families through recurrent whole genome doubling and loss of duplicate genes, as well as the nature of this loss in syntenic context via pseudogenization or DNA excision. The other focus is on phylogeny-based ancestral genome reconstruction at the level of gene order within whole chromosomes. Whole-genome doubling, tripling or replicating to a greater degree is attested in almost all lineages of the flowering plants, recurring in the ancestry of some plants two, three or more times. When this event occurs in plant genome evolution, it sets in operation a more gradual compensatory process called fractionation, the loss of duplicate genes. We make use of branching process models of recurrent whole genome duplication and fractionation, where the parameters to be estimated are fractionation rates. In the initial development, we tested the analyses on the Malvaceae and the Solanaceae. We plan to implement a full-fledged pipeline with continued developments of our method, and survey a large number of plant families for a comparative study. Adding a syntenic dimension to the study of fractionation, we explore the mechanism of fractionation. By examining how much sequence remains after a contiguous series of genes is deleted, we find that this residue remains at a constant low level independent of how many genes are lost-there are few if any relics of the missing sequence -"pseudogenes". These must be rare or extremely transient in this context. We have already verified this in ten genomes, but plan to look at  a much wider sample of genome replication events, including very recent ones in the agricultural context.  We will also look to see how frequently apparent deletion is an artifact of gene movement out of, or into, syntenic contexts. A key part of our program is to construct a pipeline 'RACCROCHE" for the phylogeny-based inference of ancestral genomes.   The strategy of this method is to accumulate a large number of generalized gene adjacencies, not just immediate neighbours, from all the extant chromosomes, subject to a phylogenetic filter at each ancestral node of the phylogeny, as input to a maximum weight matching algorithm. The latter outputs "contigs" containing up to several hundred genes each. All pairs of these contigs are then matched against each of the chromosomes in each of the extant genomes to compile a co-occurrence score and a pairwise order. The matrix of scores is then clustered to partition the set of contigs into disjoint, ordered sets,which are naturally interpretable as the chromosomes of the ancestral genome associated with the particular node of the phylogeny. We have done tests on data from the entire monocot clade and from several rosid orders.