1. genomic analysis of regulatory network dynamics reveals large topological changes Paper Study Speaker: Cai Chunhui Sep 21, 2004

2. Introduction The dynamics of a biological network on a genomic scale is presented by way of integration of transcriptional regulatory information and gene-expression data for multiple conditions in Saccharomyces cerevisiae. SANDY is developed as a new approach for the statistical analysis of network dynamics, combining well-known global topological measures, local motifs and newly derived statistics.

3. Main Work Integrate gene-expression data for the following five conditions: cell cycle, sporulation, diauxic shift, DNA damage and stress response. Fig. 1 represents the first dynamic view of a genome-scale network: the sub-networks active under different cellular conditions and standard statistics under different conditions

5. Main Work The follow-on statistics in SANDY (Fig. 2) indicates several characteristics of the regulatory system: scale-free, hubs would be invariant features of the network across conditions and the combinatorial transcription factor usage while not the individual one seems to be the key of the regulation of a condition.

8. Result SANDY presents an approach to examine biological network dynamics. In refocusing to a dynamic perspective, the author uncover substantial topological changes in network structure, and capture the essence of the transcriptional regulatory data in a new way.

9. Further Study future experiments can be used to determine condition-specific interactions directly. Many of the concepts introduced could be readily transferred to other types of biological networks and complex sub-systems in multicellular organisms (such as those directing the circadian cycle and cellular development).

10. SANDY SANDY (Statistical Analysis of Network Dynamics) is spread into three parts: Well-known statistics (global topological measures, local network motifs). Newly-derived follow-on statistics (hub usage, interchange index, TF usage). Statistical validation with randomly simulated networks

11. Good Point (Local network motifs) There are three motifs which show the precise inter-connections between a small number of TFs and target genes We are very interested in the local network motifs, and thus find out the way in which they are identified.

12. Good Point (Local network motifs) In order to identify the motifs, the author constructed a pair of affinity matrices A and B. Matrix A contained binary entries Aij where a 1 indicated a regulatory interaction from TF j to target gene i. Matrix B was a sub-matrix of A, containing only the rows corresponding to target genes that are TFs themselves. Nodes and edges can be part of more than one motif.