Biological functions mostly emerge from various interactions between genes, rather than a single gene. As the sum of the molecular interactions comprises cellular networks of enormous complexity, the studies of network topology have greatly contribute...
Biological functions mostly emerge from various interactions between genes, rather than a single gene. As the sum of the molecular interactions comprises cellular networks of enormous complexity, the studies of network topology have greatly contributed for the comprehensive understanding of how the cellular systems are functionally organized. However, the topology-based approaches have an important limitation of lacking the variety of biological interactions. Here I systematically investigated link properties in protein-protein interaction (PPI) networks through bioinformatics analyses, which incorporate biological distinction of protein interactions and apparatus of network theory.
First, I discovered that strong domain-domain interactions (DDIs) tend to connect proteins within a same biological module, whereas weak domain-linear motif interactions (DLIs) are likely to connect different biological modules. I also demonstrated that the network-based identification of functional modules can be significantly improved by taking into account those molecular characteristics of protein interactions. In addition, this division of molecular labor seemed essential for sustaining modularity in metazoan species with increased complexity than other eukaryotic species. I discovered that weak DLIs
compensated for the attenuation in module boundaries, which had been weakened by the increase of between-module interactions in the cellular systems of greater complexity.
Second, I discovered that link clustering as an indicator of contextual essential genes (EGs) that are non-central in PPI networks. In various human and yeast PPI networks, I found that 29 to 47% of EGs were better characterized by link clustering than by centrality. Importantly, such non-central EGs were prone to change their essentiality across different human cell lines and between species, accompanied with their intermediate level of gene expression and evolutionary conservation between central EGs and non-EGs. In addition, those EGs exhibited significant impact on communities at lower hierarchical levels, suggesting that link clustering is associated with contextual essentiality as it depicts locally pivotal nodes in network structure.
Finally, I developed a link-centric approach that constructs state-specific networks compatible with varying cell lines. For a given gene expression profile, the approach integrates several distinct networks selectively by omitting links that are irrelevant to the expression profile. Across 390 human cancer-cell lines, I demonstrated that the state-specific networks gave rise to the gene clusters of significant knock-out effects in the cell lines, indicating that the approach described functionally active forms. In addition, those gene clusters were topologically close to the loss-of-function mutations in the cell lines, further supporting the relevance between the resulting networks and cell-line states.