BeyondCoexpressionGeneNetworkInference.ppt

MCB CLUSTER Number of ORFs Distance from ATG (b.p.) Number of sites MIPS Functional category (# ORFs) ORFs within category DNA synthesis and replication (82) Cell cycle control and mitosis (312) Recombination and DNA repair (84) Nuclear organization (720) 23 30 11 40 N=182 Systematic Determination of Genetic Network Architecture Tavazoie et al., Nature Genetics 22, 281–285 (1999) Most motifs found are highly selective for the cluster they were found in. Can find many known binding sites for transcription factors. Also finds many novel regulatory motifs, associated with specific functional categories. 1) cluster 2) identify regulatory motifs in clustered genes 3) identify TF’s that bind to those motifs ? Gene regulation network 2: Regulatory Element Detection Using Correlation with Expression What is the contribution of each regulatory motif (or the TF that binds to that motif) to the expression level of the genes containing the motif? Given a set of known or putative regulatory motifs, identify all genes that contain the motif in their promoter region. For a single expression experiment (e.g. single point in a time series), is the presence of the motif correlated with the expression level of the genes? Perform multiple regression of (log) expression level on the presence/absence of the motifs. Plot contribution of motif throughout time series. Bussemaker et al., Nature Genetics 27, 167 – 174 (2001) : the presence of motif 1 is correlated with the expression levels of the genes in which it appears : motif 2 is not correlated with expression levels of the genes in which it appears : motif 3 is negatively correlated with expression levels of the genes in which it appears Contribution of Motifs to Expression Levels Linear Combination of Motif Contributions Find the most highly correlated motif. Determine its contribution Fi to expression level by linear regression. Subtract its contribution from the expression levels. Find the next highest correlated motif. Repeat unt