2 ms·

Genome-wide association analyses provides new insights into circadian rhythms

40 points by gwern 9y ago

yosito 9y ago
> the mean sleep timing of the 5% of individuals carrying the most "morningness" alleles was 25.1 minutes earlier than the 5% carrying the fewest Does this mean that genetics only account for 25 minutes of difference between when "morning people" and "night owls" sleep?
- gwern 9y ago
  No, that just means the subset of genetics noisily estimated by this particular method & sample can predict those extremes (ie top 5% vs bottom 5%). They don't report the full polygenic score in the paper that I see, although they note the SNP heritability is ~10% of variance (and I assume full heritability is a lot bigger since I do recall chronotype as being highly genetic).
carbocation 9y ago
It's interesting to me that they chose to limit the polygenic risk score to the SNPs at the top 292 loci, rather than using all available SNPs, which has become more common. I would guess that they chose to limit their power in favor of ease of interpretation, but it would be interesting to understand why they actually made their choice. It's also a bit surprising that being a morning person was associated with earlier age at last childbirth in women. I would have expected being a morning person to correlate with education attainment (which I don't believe they discussed) and therefore with later age at last childbirth (since education generally delays the age at which people start having children).
- auvrw 9y ago
  didn't realize this is a common method .. the correspondence with neuroanatomy is impressive. i s'p, at that level (and i mean, likely even at the proteomic level for those that're particularly studied in that area), it could be easier to have results that can guide intuition. b/c that'd surely the useful thing about these kind of results, i imagine, off the cuff, to most people in the natural sciences: providing hints for further research on causal mechanisms and perhaps a reasonably plausible story or "hypothesis" to be fancy (like ketchup) about it

40 points by gwern 9y ago

> the mean sleep timing of the 5% of individuals carrying the most "morningness" alleles was 25.1 minutes earlier than the 5% carrying the fewest Does this mean that genetics only account for 25 minutes of difference between when "morning people" and "night owls" sleep?

gwern 9y ago
No, that just means the subset of genetics noisily estimated by this particular method & sample can predict those extremes (ie top 5% vs bottom 5%). They don't report the full polygenic score in the paper that I see, although they note the SNP heritability is ~10% of variance (and I assume full heritability is a lot bigger since I do recall chronotype as being highly genetic).

carbocation 9y ago

It's interesting to me that they chose to limit the polygenic risk score to the SNPs at the top 292 loci, rather than using all available SNPs, which has become more common. I would guess that they chose to limit their power in favor of ease of interpretation, but it would be interesting to understand why they actually made their choice. It's also a bit surprising that being a morning person was associated with earlier age at last childbirth in women. I would have expected being a morning person to correlate with education attainment (which I don't believe they discussed) and therefore with later age at last childbirth (since education generally delays the age at which people start having children).

auvrw 9y ago
didn't realize this is a common method .. the correspondence with neuroanatomy is impressive. i s'p, at that level (and i mean, likely even at the proteomic level for those that're particularly studied in that area), it could be easier to have results that can guide intuition. b/c that'd surely the useful thing about these kind of results, i imagine, off the cuff, to most people in the natural sciences: providing hints for further research on causal mechanisms and perhaps a reasonably plausible story or "hypothesis" to be fancy (like ketchup) about it