IOB scientists assemble and analyze the largest interactome in humans:

The analysis included interactions among 1,077 genes coding for proteins linked to 3,133 diseases. Significantly, it showed that proteins encoded by genes that are mutated in inherited disorders were likely to interact with proteins already known to cause similar disorders. This is a novel approach to identify the new candidate genes involved in human diseases and should accelerate the research on identification of disease-causing genes.

This study also suggests that the importance of a protein cannot be determined by the number of its interacting proteins alone. This is contrary to earlier studies that suggested that more the interaction partners a protein has – the more essential it is in a cell. To study this problem, we had data on 6,014 genes in yeast, which are almost all the genes and 16,069 interactions. In addition, we obtained data on 2,284 genes in the mouse where the effect on survival of mice lacking these genes were known. In contrast, earlier studies had used only 1,870 genes and 2,240 interactions in yeast. We had a much larger dataset available for the analysis.

Our analysis on comparison of almost 25,000 human, 16,000 yeast, 5,500 worm, and 25,000 fly protein-protein interactions showed that, among these more than 70,000 links, only 16 were common to all four species. This low overlap is surprising because of a large number of interactions in each of these organisms and many biological processes are common in these organisms. We claim that this could be because of the limitations of the large-scale yeast two-hybrid method used and individual biochemical experiments must be carried out alongside high-throughput experiments if an accurate and comprehensive interaction map is to be created.

This comparison of human data with other organisms also provided us with 'connections' that had previously not been discovered in humans. We experimentally tested 9 of these 36 interactions in our laboratory and found all of them to be correct. We hope that this approach of in silicopredictions followed by experimental validation will become commonplace in the future. Importantly, this approach also allowed us to 'fill in the blanks' in cases where the circuitry in humans was missing certain components.

This analysis of human interactome will be published in the March 2006 issue of the scientific journal 'Nature Genetics,' one of the most highly cited publications. This study was led by Akhilesh Pandey, M.D., Ph.D., an Assistant Professor at the Johns Hopkins University, Baltimore, USA who established the Institute of Bioinformatics in May 2002.