27 June 2011

What's my age again?

As a bioinformatics student, most of my studies in biology thus far have focused on molecular mechanisms, genetics, and diseases. One "disease" that I find fascinating but know relatively little about, is aging. With talk of telomere lengths and "fountain of youth" pills, I find the topic of aging to be interesting, but haven't seen much of my sort of computational research applied to the subject. For this reason, I highly recommend all you data miners out there to check out a recent paper in PLoS One by Sven Bocklandt and colleagues at UCLA. The concept is fairly simple: input DNA methylation patterns (labeled with age), build model, predict age. Result: "we built a regression model that explained 73% of the variance in age, and is able to predict the age of an individual with an average accuracy of 5.2 years". The initial model was based on 34 sets of male identical twins, then validated in 60 additional people. It would be interesting to see if lifestyle factors (e.g. smoking or drinking, which can affect DNA methylation) are important covariates in their model. Kudos to all those involved for this simple and elegant yet effective study. 
Original Paper: Epigenetic Predictor of Age