Friday, 14 November 2014

Supercentenarian genomes: but are they the right ones

GenomeWeb covered a recent paper from Stuart Kim's group at Stanford University: Whole-Genome Sequencing of the World’s Oldest People. Unfortunately they did not find any longevity genes, and Kim was quoted saying "We were pretty disappointed."

There is lots of suggestion that longevity has a genetic component, but I can't help but consider that the environmental component is likely to be stronger, and this would mask the genetics. Kim's study was very small, just 17 genomes, so the chances of finding anything were equally small. But had there been low-hanging fruit this would almost certainly have been a Science paper rather than PLoS One (sorry PLoS).

Sample size matters: We run Tuesday afternoon experimental design sessions for our scientists, where they can talk with people from the Genomics and Bioinformatics core facilities. In these we often discuss sample size and replication as the two most important factors to consider. I'm sure we'd have said run more samples but that is perhaps too easy to say, especially given the hindsight of the data.

But it is not just sample size that matters, the right sample needs to be considered. It was not clear from the paper what the heritage of the supercentenarians was. Given the requirement or birth certificates and driving licences I'd guess these were all very old Americans.

Did the use of very old American women reduce the chances of finding the longevity gene? This is not possible to say without further study. However I would have thought selecting supercentenarians from a wealthy and healthy population (in global terms) is likely to have added significant bias.

So who should we sequence to find the longevity gene? I'd aim to look at very old people in the third world, my assumption is that if you've survived to a ripe old age in an  challenging environment then your genetics must be pretty good. The problem is there may not be any very strong longevity genes, longevity matters much less than fecundity so there probably been very little selection going on.

PS: If you're interested in looking at the data you can get it on Google Genomics, thanks to Nicolas for directing me to this right page.

PPS: here's a list of veveryone in the UK over 107,, only 1 bloke in the top 30!.

1 comment:

  1. I am not sure it adds additional bias to take "supercentenarians from a wealthy and healthy population". If your control population is selected with similar condition, it's fine (and the conclusion are restricted to that population). But you'd have to select people who died younger and didn't die from a disease, right?

    Here is how to access the data through Google Genomics:

    Seems easier than dbGap...