Conservation, error and dynamics in protein interaction networks.

Thesis

The availability of large scale protein interaction networks for several species has motivated many comparative studies in recent years. These studies typically employ network alignment algorithms for the task and use the sequence similarity of proteins to aid the alignment process. In this thesis I use a quantitative measure of protein functional similarity and show that the results are superior to sequence based network alignment. I present a method for module detection that combines results from network alignments with clustering measures to achieve superior results over several existing methods. Next, I address the issue of generally low conservation detected by alignments of interaction networks from model organisms. By explicitly modelling evolutionary mechanisms on pairs of networks I test the hypothesis that divergent evolution alone may be the cause. I use a distance metric based on graph summary statistics to assess the fit between experimental and simulated network alignments. Our results indicate that network evolution alone is unlikely to account for the poor quality alignments given by real data. We also find that false positives appear to affect network alignments little compared to false negatives indicating that incompleteness, not spurious links, is the major challenge for interactome-level comparisons. Finally, I focus on the comparative analysis of a subset of the interaction network related to mitosis in Yeast, Human and Fly. Manual ordering of mitosis-related functional annotations allows the study of temporal aspects of the network. I also use a Markov random field approach to infer temporal labels for unlabelled proteins. Sequence based network alignment of the mitotic networks in the three species finds little conservation despite the proteins being functionally very similar. Further investigation suggests a fuzzy relationship between protein sequence and function that may have implications for future network alignment studies.

Authors: Ali, W

• Publication date: 2011
• DOI: 10.5287/ora-xk874a1po
• Type of award: DPhil
• Level of award: Doctoral
• Awarding institution: Oxford University, UK
• Language: English
• Keywords: protein interactions; network alignment
• Subjects: Mathematical biology