Social Network Analysis of DNA Microarray Data
Jung Hun Ohn1, Tae Su Chung2, Jihoon Kim, Mingoo Kim, Jihun Kim, Hye Won Lee, Ji Yeon Park, Ju Han Kim
1jhohn2@snu.ac.kr, SNUBI Seoul National University Biomedical Informatics; 2epiai@korea.com, SNUBI Seoul National University Biomedical Informatics
Cellular processes, from mechanical view, can be regarded as sequential arrangement of protein actions to make essential product for the survival of the cell. Only with individual components of a process(i.e. genes or proteins), however, the purpose or intention of a cellular process may not be fully explained. Study of relationships between each components can give us better understanding of cellular process. We applied the wisdom of social analogy to find out the relations of gene expression. The social network analysis explains the property of individual genes through their social groups and relationships.
One-mode and two-mode analysis with the analogy of genes as actors and experimental conditions as events represented in an affiliation matrix identified the social ‘stars’and the network structures from the Rosetta Compendium dataset with 300 diverse mutations and chemical treatments in S. cerevisiae. Bipartite matrix and bipartite graph were constructed. We measured the rates of participation, size of events, and the centrality indices such as node degree, closeness, and betweenness centralities. Group centralization measures, clustering structure and core-periphery structure were determined in company with MIPS functional classification to explore the genome-wide interaction structure.
Regarding genes as actors, conditions as events and the genetic network topology as variety of centrality and relatedness indices, social network analysis was performed on large-scale gene expression profile. The analysis demonstrated some important features such as core-peripheral players and significant intermediary actors that may be critical for the control of the system and useful for the development of valuable therapeutic substances.