This preview shows pages 1–3. Sign up to view the full content.
This preview has intentionally blurred sections. Sign up to view the full version.View Full Document
Unformatted text preview: 1 NETWORKING PATHWAYS UNVEILS ASSOCIATION BETWEEN OBESITY AND NON-INSULIN DEPENDENT DIABETES MELLITUS * HAIYAN HU School of Informatics and Center for Computational Biology and Bioinformatics, Indiana University, 410 West 10th street, Suite 5000, Indianapolis, IN 46202, USA XIAOMAN LI Division of Biostatistics, Indiana University, 410 West 10th street, Suite 5000, Indianapolis, IN 46202, USA Genetic related health problems are often interrelated. Current practices to establish associations between diseases are expensive and rarely can reflect underlying molecular mechanisms. We propose a general framework to associate diseases by networking pathways. By applying our method on association study of non-insulin dependent diabetes mellitus (NIDDM) and obesity, we demonstrate that our method can both identify signature pathways for each disease and establish valid association of two diseases. 1. Introduction Many diseases are interrelated. Obesity, diabetes, insulin resistance, hypertension are just a few examples. Instead of being attributed to a specific gene, these diseases are often caused by interaction among multiple genes or between genes and environment, and thus are often classified as multifactorial disease or complex disease. Great effort has been put on association studies, such as case control studies and cohort studies, to discover the potential relation between multiple disease conditions in human. Although such association studies can often produce very important information, they are either not very reliable or not efficient in terms of time and money. For example, of two large American Cancer Society cohorts, Cancer Prevention Study I (CPS-I; enrolled in 1959 and followed through 1972) and Cancer Prevention Study II (CPS-II; enrolled in 1982 and followed through 1996), one shows association of height with prostate cancer, the other does not . Most importantly, from such association studies on complex diseases involving genetic factors, no matter how significant the identified associations are statistically, researchers usually cannot gain much insight of the underlying molecular mechanisms. Thus, * This work is partially supported by Indiana Genomics Initiative (INGEN), by Showalter Trust award and by R01HG004359 from NHGRI. Corresponding authors. Pacific Symposium on Biocomputing 13:255-266(2008) efficient methods are urgently needed to identify disease associations at the molecular level. Microarray experiments have been a very popular tool for disease study. From microarray data, gene expression signatures that can distinguish a disease phenotype from another have often been identified by implementing analytical techniques such as differential test. However, in complex diseases like cancer, it is not the individual genes but the interaction between many genes and the interaction between many genes and environment that are responsible for a certain physiological process. Therefore, dozens of suspicious genes included in certain physiological process....
View Full Document
- Spring '09