Towards Personalized Interactomes

This thesis presents a preliminary study to evaluate the ability of our sequence-based protein-protein interaction prediction tool (PIPE) to detect changes in the interactome caused by non-synonymous Single Nucleotide Polymorphisms (SNPs). Sequence-based PPI prediction is not sensitive to SNP-induced interactome changes, however sequence-based PPI interaction site prediction can be used to extract information regarding interactome changes. PIPE on its own does not perform well on detecting the effects of SNPs. However, this lack of sensitivity is not limited to PIPE, but likely affects all sequence-based methods. Using the interaction site prediction feature of PIPE, PIPE-Sites, a number of interactions are identified for which there is reason to believe that they may, in fact, be affected by SNPs. To examine the effect of co-occuring SNPs, genotypes are extracted from the 1000 Genomes initiative. It appears that PIPE-Sites is able to identify subsets of interactions likely affected by particular genotypes.