Integrative -omics for the Systems Biology of Complex Phenotypes

Wednesday, February 29, 2012 - 2:00pm - 2:15pm
Keller 3-180
Mehmet Koyuturk (Case Western Reserve University)
In the study of biological systems, many phenotypes, i.e., observed
differences between the behavior of different organisms, are complex -
that is, they are based on a set of complex interactions between
multiple genetic and environmental factors. Genomic association
studies provide significant information on the genetic bases of
complex phenotypes, however so far have defined a limited set of
phenotypes. Large scale monitoring of gene expression reveals
underlying cellular mechanisms, however, it is limited in capturing the
abundance and activity of functional proteins. Protein expression data
provides more reliable information on function, but with limited
coverage. Protein-protein interactions (PPI) highlight functional
relationships between proteins, but PPI data is highly noisy,
incomplete, and static. In this talk, we investigate how these useful,
yet limited sources of biological data can be integrated through
innovative use of computational approaches to gain insights on the
molecular mechanisms of complex phenotypes. We introduce several
algorithmic problems that stem from these applications, including (i)
network-based prioritization of candidate disease genes, (ii)
integration of protein and gene expression data to identify
dysregulated subnetworks in cancer, and (iii) use of subnetwork
markers in prediction of metastasis. Experimental results on various
public and dedicated datasets show that such integrative approaches
can provide significant insights into the systems biology of complex