A complex disease like cancer is hardly caused by one gene or one protein singly. It is usually caused by the perturbation of
the network formed by several genes or proteins. In the last decade several research teams have attempted to construct interaction maps
of genes and proteins either experimentally or reverse engineer interaction maps using computational techniques. These networks were
usually created under a certain condition such as an environmental condition, a particular disease, or a specific tissue type. Lately, however,
there has been greater emphasis on finding the differential structure of the existing network topology under a novel condition or disease
status to elucidate the perturbation in a biological system. In this review/tutorial article we briefly mention some of the research done
in this area; we mainly illustrate the computational/statistical methods developed by our team in recent years for differential network
analysis using publicly available gene expression data collected from a well known cancer study. This data includes a group of patients
with acute lymphoblastic leukemia and a group with acute myeloid leukemia. In particular, we describe the statistical tests to detect the
change in the network topology based on connectivity scores which measure the association or interaction between pairs of genes. The
tests under various scores are applied to this data set to perform a differential network analysis on gene expression for human leukemia.
We believe that, in the future, differential network analysis will be a standard way to view the changes in gene expression and protein expression
data globally and these types of tests could be useful in analyzing the complex differential signatures.
Keywords: Differential network analysis, human acute leukemia, permutation test, genetic modules.
Rights & PermissionsPrintExport