Charité - Universitätsmedizin Berlin researchers have developed a computational method to accelerate the identification of disease genes, as Köhler and colleagues report in the latest edition of the American Journal of Human Genetics*. The identification of disease genes is essential for the elucidation of genetic disorders and for their treatment. Since the initial identification of a disease gene in 1988 (the CFTR gene mutations in which cause cystic fibrosis), about 2500 further disease genes have been identified. Scientists estimate that up to 3500 further disease genes have yet to be discovered.
At present, the identification of genes in positional cloning projects can be time-consuming and expensive. At first, a region in one of the 24 human chromosomes is identified by linkage analysis or association studies. In general, such regions may contain 100 to 300 genes or more. If one does not know which gene is responsible for the disease, the search for the causative gene may involve sequencing one gene after another, which may entail years of work and high costs.
The new method developed by the Computational Biology and Bioinformatics Group of the Institute of Medical Genetics (http://compbio.charite.de) aims to identify the most likely candidate genes with computational methods. The researchers analysed the interactions between different genes within the protein interactome. They observed that genes associated with phenotypically similar diseases tend to be tightly interconnected to one another within the cellular network. They adapted the "random walk" method in order to traverse all paths emanating from known disease genes.to candidate genes. The results of testing on 110 diseases showed clearly better results than previous methods that use protein interactions to predict disease genes.
Abstract available online: http://www.ajhg.org/AJHG/latestarticles
Author contact: Peter N. Robinson
Message posted by: Frank S. Zollmann