Biological Regulatory Network Discovery

Matthew Hudson, Assistant Professor of Genomics, UIUC

The model plant Arabidopsis undergoes a complete change in developmental and metabolic direction when the seedling first detects light. This series of changes is initiated by a quantum event when a photon is captured by a photoreceptor, phytochrome A, and leads by a transcriptional cascade to wholesale changes in the transcription of the genome. We have developed an efficient algorithm to identify over-represented DNA motifs, which are a signature that can be used to discover the components of this regulatory system. Using this approach, we have identified new cis-regulatory DNA motifs using Arabidopsis transcriptional profiling and genomic sequence data. These motifs are capable of conferring light-inducibility on a reporter gene in vivo. We have used the output of fourteen motifs from the motif discovery software to develop a support vector machine approach to predicting the expression of Arabidopsis genes from the sequence of their promoter regions. Using this model, we are able to predict the behavior of a large number of Arabidpsis light-induced genes in silico, showing that the promoter region is the primary determinant of light induced transcription.