OpenPsych forums

Full Version: Ethnic Differences in the UK
You're currently viewing a stripped down version of our content. View the full version with proper formatting.
Pages: 1 2 3 4 5 6 7 8 9 10 11 12 13
(2015-Nov-18, 03:20:27)Zoidberg Wrote: [ -> ]I said I wouldn't post any more about epigenetics in this topic, but I changed my mind.

They are attacking the entire genetic model of heritability studies and these additive SNP correlations, with epigenetics as their main proof against it. They are getting published on major scientific journals now and also being cited by potential reviewers of the work here.

Heres a correspondence.
Attack:
http://onlinelibrary.wiley.com/doi/10.11...12060/full
Defence:
http://onlinelibrary.wiley.com/doi/10.11...9/abstract

The defence failed by the way, since they try to play off epigenetics as GxE and rGE, which it isn't.

I knew this was coming, you are lucky they don't have the silver bullet... yet.

What does it matter? Firstly, biometric modeling can incorporate epigenetics as a variance component. So pointing out that epigenetics is a possible factor does not decrease the utility of the method. In fact, it increases it. If researchers want to show the epigenetic factors account for a substantial portion of the variance in a given trait they should employ a biometric model instead of pointing to local associations.

Second, present methodologies can partition variance in a way that can exclude possible epigenetic effects. For example GCTA can not be so confounded, nor can kinship estimates of AM and dominance. Nor can estimates of share environment based on unrelated sibs reared together. Nor can estimates of measurement error. Epigenetic effects could explain some portion of unshared environment -- it would just count as another form of intrauterine effect. But so what? And it could explain some residual H^2 as H^2 - (h^2SNP + H^2D + H^2AM) but for traits like g, there is not much here. For other traits e.g., personality, that's another issue. But Hereditarians don't usually focus on these -- because there aren't large differences between the groups of interest! Indeed, given this, epigenetics could salvage some hereditarian positions, by offering an explanation for the lacks of apparent differences i.e., a masking effect.

Third, whole genome comparisons will substantially narrow the uncertainty since they will capture rare variants, which are thought by many to account for the "missing H^2kinship".

IMO, you're being silly.

(Also, as I have noted numerous times, epigenetic differences are perfectly consistent with a racialist model since, for one, early such models prior to Darwin (who adopted a Lamarckian frame) were mostly epigenetic, at least when races were conceptualized as intraspecfic divisions.)
[quote='Chuck' pid='3683' dateline='1448033860']
[quote='Jm8' pid='3681' dateline='1447780919']


"There results are not very informative since the studies looked at different trait associated alleles. Thus, there was an apples to oranges comparison. The African-European comparisons concerned primarily Asthma related alleles, which show low replicability across all populations. You would want to compare the effect of alleles associated with the same traits, for example height, to get a sense if associations that replicated between Asians and Europeans also did between European and Africans."

The study appears to claim that the alleles for those traits correlated well between Asians and Europeans but less so for Africans.

"Two reviews found substantial cross-validity for the Eurasian population"
...~100% of SNPs replicate in other European samples when accounting for statistical power, ~80% in East Asian samples but only ~10% in the African American sample (not adjusted for statistical power, which was ~60% on average). There were fairly few GWAS for AAs however, so some caution is needed in interpreting the number."
https://thewinnower.com/papers/2735-poly...fer-method

"Regarding the alleles for cognitive ability, the association -- at least with regards to cognitive scores -- has been replicated in an East Asian sample and in a nationally representative African American one."

Which is the African American study. Do you have a link?
The same study I linked also seems to find:

"LOW CROSS-VALIDITY OF GWAS BETAS AND POLYGENIC SCORES FOR EDUCATIONAL ATTAINMENT IN AAS(African Americans:my parenthesis) "

(title of section 6)
(2015-Nov-21, 22:40:34)Jm8 Wrote: [ -> ]The same study I linked also seems to find:

"LOW CROSS-VALIDITY OF GWAS BETAS AND POLYGENIC SCORES FOR EDUCATIONAL ATTAINMENT IN AAS(African Americans:my parenthesis) "

(title of section 6)

You realize that you are quoting my paper, right? :)
(2015-Nov-20, 18:04:02)Chuck Wrote: [ -> ]What does it matter? Firstly, biometric modeling can incorporate epigenetics as a variance component. So pointing out that epigenetics is a possible factor does not decrease the utility of the method. In fact, it increases it. If researchers want to show the epigenetic factors account for a substantial portion of the variance in a given trait they should employ a biometric model instead of pointing to local associations.

Second, present methodologies can partition variance in a way that can exclude possible epigenetic effects. For example GCTA can not be so confounded, nor can kinship estimates of AM and dominance. Nor can estimates of share environment based on unrelated sibs reared together. Nor can estimates of measurement error. Epigenetic effects could explain some portion of unshared environment -- it would just count as another form of intrauterine effect. But so what? And it could explain some residual H^2 as H^2 - (h^2SNP + H^2D + H^2AM) but for traits like g, there is not much here. For other traits e.g., personality, that's another issue. But Hereditarians don't usually focus on these -- because there aren't large differences between the groups of interest! Indeed, given this, epigenetics could salvage some hereditarian positions, by offering an explanation for the lacks of apparent differences i.e., a masking effect.

Third, whole genome comparisons will substantially narrow the uncertainty since they will capture rare variants, which are thought by many to account for the "missing H^2kinship".

IMO, you're being silly.

(Also, as I have noted numerous times, epigenetic differences are perfectly consistent with a racialist model since, for one, early such models prior to Darwin (who adopted a Lamarckian frame) were mostly epigenetic, at least when races were conceptualized as intraspecfic divisions.)

It looks like whole genome sequencing could be somewhat unnecessary as imputation seems to be fairly useful.

http://www.nature.com/ng/journal/v47/n10....3390.html

Still, there is going to be some rare variants that can only be seen in full genome sequencing.

Full genome sequencing opens up for some new genetic possibilities. For instance, MZ twins are not actually genetically identical, but may have important de novo mutations. These de novo mutations can arise at any point during development, so one may have to sequence brain matter to see them. One can essentially do a GCTA on MZ pairs. Do the MZ pairs with more mutations between them (in their brain tissue) differ more in cognitive ability? (Controlling for age.)
(2015-Nov-20, 17:37:40)Chuck Wrote: [ -> ]
(2015-Nov-17, 19:21:59)Jm8 Wrote: [ -> ]The findings related to the relatively poor, but non-zero cross-validity of GWAS betas between European and African samples throw some doubt on the SNP evidence found by Piffer in his studies of the population/country IQ and cognitive ability SNP factors (29). If the betas for the SNPs identified in European sample GWAS do not work well as predictors for Africans, they would be equally unsuitable for estimating mean genotypic cognitive ability from SNP frequencies. Thus, further research is needed to more precisely estimate the cross-racial validity of GWAS betas, especially with regards to African vs. Eurasian samples."

There results are not very informative since the studies looked at different trait associated alleles. Thus, there was an apples to oranges comparison. The African-European comparisons concerned primarily Asthma related alleles, which show low replicability across all populations. You would want to compare the effect of alleles associated with the same traits, for example height, to get a sense if associations that replicated between Asians and Europeans also did between European and Africans. (I pointed this out to Emil, but he neglected to comment on the issue.)

Regarding the alleles for cognitive ability, the association -- at least with regards to cognitive scores -- has been replicated in an East Asian sample and in a nationally representative African American one.

I forgot about/didn't notice that.

Their approach to estimating cross-racial validity is silly. They use a hits-based (dichotomous thinking) approach, whereas they should correlate the betas of all the SNPs from the largest GWAS of each group (continuous approach).
(2015-Nov-21, 23:28:51)Emil Wrote: [ -> ]
(2015-Nov-21, 22:40:34)Jm8 Wrote: [ -> ]The same study I linked also seems to find:

"LOW CROSS-VALIDITY OF GWAS BETAS AND POLYGENIC SCORES FOR EDUCATIONAL ATTAINMENT IN AAS(African Americans:my parenthesis) "

(title of section 6)

You realize that you are quoting my paper, right? :)

Yes.
(2015-Nov-21, 23:35:56)Emil Wrote: [ -> ]Do the MZ pairs with more mutations between them (in their brain tissue) differ more in cognitive ability? (Controlling for age.)

This has been done, with negative results. "CNV concordance rates were compared between the different sources of DNA, and gene-enrichment association analyses were conducted for thought problems (TP) and attention problems (AP) using CNVs concordant within MZ pairs. he gene-enrichment analyses on concordant CNVs showed no significant associations between CNVs overlapping with genes involved in neuronal processes and TP or AP after accounting for the source of DNA."

This is another blow to the mutation-load theory of intelligence, especially after the finding that paternal age isn't related to offspring IQ.
Twin Research and Human Genetics: http://dx.doi.org/10.1017/thg.2014.86
(2015-Nov-21, 23:38:50)Jm8 Wrote: [ -> ]
(2015-Nov-21, 23:28:51)Emil Wrote: [ -> ]
(2015-Nov-21, 22:40:34)Jm8 Wrote: [ -> ]The same study I linked also seems to find:

"LOW CROSS-VALIDITY OF GWAS BETAS AND POLYGENIC SCORES FOR EDUCATIONAL ATTAINMENT IN AAS(African Americans:my parenthesis) "

(title of section 6)

You realize that you are quoting my paper, right? :)

Yes.

It's unfortunate that such a statement is becoming so popular among anti-hbd folks, because in my opinion it is wrong. Differences in LD should simply reduce the observed allele frequencies differences between populations. There is no reason why these should be smaller than they appear. Correction for attenuation due to different LD should actually make the results observed by Piffer even stronger.
Edit: by observed I mean "at the sampled SNPs"..by real I meant "at the real CAUSAL variants". So correction for attenuation will increase race differences at causal variants compared to differences at the observed variants (which are in LD with the causal variants)".
(2015-Nov-23, 20:50:51)Duxide Wrote: [ -> ]
(2015-Nov-21, 23:35:56)Emil Wrote: [ -> ]Do the MZ pairs with more mutations between them (in their brain tissue) differ more in cognitive ability? (Controlling for age.)

This has been done, with negative results. "CNV concordance rates were compared between the different sources of DNA, and gene-enrichment association analyses were conducted for thought problems (TP) and attention problems (AP) using CNVs concordant within MZ pairs. he gene-enrichment analyses on concordant CNVs showed no significant associations between CNVs overlapping with genes involved in neuronal processes and TP or AP after accounting for the source of DNA."

This is another blow to the mutation-load theory of intelligence, especially after the finding that paternal age isn't related to offspring IQ.
Twin Research and Human Genetics: http://dx.doi.org/10.1017/thg.2014.86

Good find, tho not completely convincing.

By paternal age finding, my guess is that you mean the Arslan et al paper? http://journals.plos.org/plosone/article...ne.0090097

Look over their analyses. They report detailed R output. See files under (3) here https://osf.io/wlrzf/wiki/home/.

Model 2 is the right one to look at IMO. In Model 3, they over-correct because they use birth weight, birth order and birth complications as controls. Paternal age is related to birth orders, and mutations are plausible related to birth weight and birth complications. Thus, controlling for these is over-correcting because it indirectly controls for the independent variable of interest. This is similar to the sociologist's fallacy.
Pages: 1 2 3 4 5 6 7 8 9 10 11 12 13