> the effect is subtle, explaining just 1 percent of a person’s depression risk.
That seems like noise to me. Especially in such a large study that attempted to cross-correlate (apparently) hundreds of factors, I would require a lot more evidence not to discard this as the jelly bean effect (https://xkcd.com/882/).
Now I haven't read the paper, but based on this article, I dare say it would be irresponsible to label this a finding.
In addition, even if there is a real effect here, it is so small, and the interactions so complex, that it's probably useless.
For example, even if this neanderthal gene happens to causally affect depression by 1%, that pathway might be something like
gene => slightly wider shoulders => that body structure is slightly more or less likely to lead to certain life outcomes, like marrying early or getting promoted or getting targeted by pickpockets => change in depression risk
As a result, the gene might currently lead to a 1% increase in depression risk, but in a few years it might do the opposite, if things change in the surrounding culture and those intermediary steps behave a little differently.
1% is a lot for a complex trait like depression. All genetics combined predicted/determines ~37%, and the part specifically attributable to SNPs is ~32% (http://www.ncbi.nlm.nih.gov/pmc/articles/PMC3404250/). So that one variant is explaining more like 3% of the genetics; another 32 such variants and about all the possible genetics is known! And they could potentially lead to major insights by looking at what the Neanderthal variant does differently from the regular Homo sapiens variant does; a recent example of this was synaptic pruning & schizophrenia.
They used a validation set of 14,000 people. Any correlations found in the first group is checked against the second group to make sure it's a real correlation.
Psychiatric disorders are commonly found to be influenced by a multitude of genes (SNPs, CNVs, etc.), often dozens or even hundreds. Each genetic factor accounts for only a small part of the difference between affected and normal groups. Frequently, perhaps mostly, the proportion is <5%, often as low as <=1%. As the article says, there are many genetic and environmental factors involved in complex conditions like depression or schizophrenia, so these low levels of effect are unsurprising.
It's also discouraging that they support their finding by saying "The significant replicated association of Neandertal SNPs with mood disorders, in particular depression, is intriguing because Neandertal alleles are enriched near genes associated with long-term depression (5)".
"Long-term depression" (LTD) has absolutely nothing to do with "depression". LTD is a mechanism of synaptic plasticity, where the strength of the synapse is reduced ("depressed").
"They started with 13,700 people from the eMERGE Network, and looked for associations between 135,000 Neanderthal genetic variants and 1,689 different traits."
Does anyone know how they compensated the statistics to control for the p-hacking?
I took a brief look at the paper scanning for multiple hypothesis testing corrections.
"Neandertal variants have been hypothesized
to influence many phenotypes in AMHs, including
lipid metabolism, immunity, depression,
digestion, hair, and skin, on the basis of the enrichment
of Neandertal variants in regions of
the genome relevant to these traits. Accordingly, we first tested these hypotheses
using genome-wide complex trait analysis (GCTA)
to estimate the phenotypic risk explained by 1495 genotyped common (minor allele frequency
> 1%) Neandertal SNPs for a set of 46 highprevalence
phenotypes from the hypothesized
categories, using age, sex, and eMERGE site as
covariates."
When testing these 46 traits they controlled the FDR < 0.05 (false discovery rate). Then they replicated the associations in an independent sample.
The effect sizes are pretty small (1-2%), so having the explanatory variants won't have much bearing whether or not you'll get these complex traits.
You missed the next sentence: "They then checked any links they found against a second group of 14,700 eMERGE volunteers."
So if any correlations were found, they basically did another study to replicate it. And that's a big sample size too. Many studies are published with far weaker correlations.
That's true, but the effect size here is also tiny - 1%. It's plausible it's statistically significant but completely meaningless.
It's possible this study is novel mainly for the methodology, which as reported is the first wide-scale test of its nature. The media of course focus on the correlations because they are more sensational - it gets people thinking "am I depressed because I'm part neanderthal?"
Well the base rate of being depressed is low, so an increase of 1% is kind of significant. If the base rate is only 1%, then a 1% increase means you are twice as likely to be depressed!
I agree it's meaningless, in the sense that we have no idea what the cause is. It could be indirect, it could be random rather than anything special about neanderthals, it could be confounded with ethnicities that have higher neanderthal DNA, etc.
I think it's a 1% explanatory effect of the total variance. That means that 1% of the variance in the population (not the size of the effect) is explainable by this factor.
(Also, the base rate of depression is far more than 1%, see e.g. http://tech.mit.edu/V120/N3/dep1.3n.html - it's around 17% for just a major depressive episode.)
Even Darwin had noticed this, when he described the differences between groups which (unbeknownst to him) had Neadenthal origins and those who don't.
"There is however no doubt that the various races when carefully compared and measured differ much from each other as in the texture of the hair, the relative proportions of all parts of the body, the capacity of the lungs the form, and capacity of the skull, and even in the convolutions of the brain. But it would be an endless task to specify the numerous points of difference. The races differ also in constitution in acclimatisation and in liability to certain diseases. Their mental characteristics are likewise very distinct chiefly as it would appear in their emotional but partly in their intellectual faculties. Every one who has had the opportunity of comparison must have been struck with the contrast between the taciturn, even morose, aborigines of S. America and the light-hearted talkative negroes."
And to think, some fundamentalists want to ban discussion of Darwinian evolution.
You'll be depressed too... Once the transhumanists are successful in their plot to rule the entire planet, while they constantly insist on using "Homo Sapien Sapiens" as a colloquial derogatory term for "those intellectually-challenged violent brutes" ;)
That seems like noise to me. Especially in such a large study that attempted to cross-correlate (apparently) hundreds of factors, I would require a lot more evidence not to discard this as the jelly bean effect (https://xkcd.com/882/).
Now I haven't read the paper, but based on this article, I dare say it would be irresponsible to label this a finding.