How to assess non-randomness of viral integration in the genome

Question

[Nicolas Rosewick]

Hi,

I've three species (mouse, rat and bovine) where N=100, M=150 and O=130 viral integrations occurred respectively. The virus is not exactly the same in the three species but is very close functionally.

Each species harbor different number of chromosomes, thus different genomic organizations (gene position, etc...).

Question : How can I assess the non-randomness of viral integration using my set of 100+150+130 viral integrations.

The difficulty here is working with three different species. An idea was to remap all integration sites onto one species by looking to homologous region but not every integration sites are within region that are found in genomic regions of the other species...

Any idea?

Thanks

Answer 1

[Michael]

I think by synteny could be an approach. Create gene sets from genes close to the integration site, and look for over-representation of orthologs, GO terms, etc, by using standard over-representation analysis. Check for enrichment of other genomic features (methylation, CpG, GC content) in the proximity of the integration site.

[Nicolas Rosewick]

That was something I had also in mind. Take the closest genes up and downstream of each integration and check for enrichment of ortholog genes.

[Michael]

How to define the neighborhood? With GSEA you could do a comparison of orthologs ranked by distance. according to this article is depends on open chromatin though: http://www.ncbi.nlm.nih.gov/pubmed/15168737

[Nicolas Rosewick]

Integration depends on open chromatin -> ok. But proliferative advantage might depend on specific integration sites or regions. Thus "old / long-life" clones might possess integration in recurrent regions.