Home

About Us: Our Halo Research

ISB Halo Group

Educational Outreach

Publications

Raw Data

Genomic Sequence

Regulatory Network

Physical/Functional Network

Related Websites

Picture Gallery

Guest Book

        Our attempt to construct gene regulatory networks for Halobacterium sp. is currently based primarily on microarray experiments, whereas the metabolic and physical interaction/function networks are based primarily on protein interactions or related functions inferred from other organisms. We used operon prediction using previously published methods along with steady state microarray data to cluster genes into strictly co-regulated groups. For each co-regulated cluster we find the model that best describes its transcript profile as a function of the transcript profiles for other genes or co-regulated gene clusters in the network. This can be achieved using Boolean, linear and generalized linear functional forms (models) to describe how genes change as a function of transcription activators and repressors

      We have recently expanded on our previous work to include the linear and generalized linear models for gene activation/repression. We reduce the problem of deconvoluting regulatory influences from microarray experiments to a problem of linear (generalized linear) regression and subset selection. Using regression methods we have inferred a large number of probable regulatory influences between clusters of co-regulated genes.

      The interactions within the regulatory influence network has been limited to outgoing edges from clusters containing transcription factors and DNA binding proteins identified on the basis of Pfam (Pfam is a large collection of multiple sequence alignments and hidden Markov models covering many common protein domains and families) and COG classifications and Rosetta ab initio structure predictions.