Thursday, July 05, 2007

A top-down approach to genetic networks   posted by p-ter @ 7/05/2007 08:08:00 PM

One of the interesting findings to come from the recent burst of genome-wide association studies is that many seemingly disparate phenotypes share some genetic pathways. I don't think many people would have a priori considered the possibility of a genetic link between prostate cancer and type II diabetes, yet that's what the data suggest. Other links are somewhat more predictable-- type I diabetes and Crohn's disease both have an autoimmune component, so a genetic link might have been expected. And cornary heart disease and type II diabetes share some envronmental risk factors, so perhaps it's to be expected that similar genetic networks play a role in the two.

These genetic links have all been uncovered by classic reductionist methods-- find the molecular variation that predisposes to disease 1, find the molecular variation that predisposes to disease 2, and compare the two sets. It's simple, and it works.

However, a clever new paper takes a different approach:
Geneticists and epidemiologists often observe that certain hereditary disorders cooccur in individual patients significantly more (or significantly less) frequently than expected, suggesting there is a genetic variation that predisposes its bearer to multiple disorders, or that protects against some disorders while predisposing to others. We suggest that, by using a large number of phenotypic observations about multiple disorders and an appropriate statistical model, we can infer genetic overlaps between phenotypes. Our proof-of-concept analysis of 1.5 million patient records and 161 disorders indicates that disease phenotypes form a highly connected network of strong pairwise correlations. Our modeling approach, under appropriate assumptions, allows us to estimate from these correlations the size of putative genetic overlaps.
This is data mining at its finest.

I'll admit I didn't go through all of the hundreds of pages of supplementary material, but this is fascinating stuff. The authors seem to be particularly interested in autism, which they find correlates with a number of neurological disorders, but also bacterial and viral infections and autoimmune disease:
Our estimated significant overlap between autism and tuberculosis may indicate that both diseases are associated with genetic changes weakening the immune system
Or consider the overlap between bipolar disorder and breast cancer:
Although the competitive genetic overlap between bipolar disorder and female breast cancer has not been reported, there is recent indirect evidence that supports it: a well-established breast cancer drug, tamoxifen, was recently discovered to be effective in treating symptoms of bipolar disorder.
The amount of data generated by the medical community each day is staggering, and as genetic information gets cheaper, it will increasingly be a part of that data. Half the game is knowing what to look for.

Labels: ,