Share this page!

Enter URL or ID to Unroll

You can paste full URL like: https://x.com/threadreaderapp/status/1644127596119195649
or just the ID like: 1644127596119195649

How to get URL link on X (Twitter) App

  1. On the Twitter thread, click on or icon on the bottom
  2. Click again on or Share Via icon
  3. Click on Copy Link to Tweet
  4. Paste it above and click "Unroll Thread"!
  5. More info at Twitter Help
Sasha Gusev Profile picture
Sasha Gusev
@SashaGusevPosts
Statistical geneticist. Associate Prof at @DanaFarber | @harvardmed | @DFCIPopSci
John Smith⚛ (ananthropocentric purposivism) 🌎 Profile picture Maria Francis Palafox Profile picture Victor Fernandes Profile picture 4 subscribed
May 12 • 21 tweets • 9 min read
I've written the first part of a chapter on the heritability of IQ scores. Focusing on what IQ is attempting to measure. I highlight multiple paradoxical findings demonstrating IQ is not just "one innate thing".



I'll summarize the key points here. 🧵 gusevlab.org/projects/hsq/#…
Image First, a few reasons to write this. 1) The online IQ discourse is completely deranged. 2) IQists regularly invoke molecular heritability as evidence for classic behavioral genetics findings while ignoring the glaring differences (ex: from books by Ritchie and Haier/Colom/Hunt).
Image
Image
Apr 30 • 15 tweets • 6 min read
It pains me to see facile critiques of GWAS on here from our clinical/biostats friends while the many actually good reasons to be critical of GWAS get little attention. So here's a thread on what GWAS does, what critics get wrong, and where GWAS is genuinely still lacking. 🧵: Here’s an example of what I’m talking about from Frank Harrell’s otherwise excellent critique of bad biomarker analysis []. This gets GWAS completely wrong. Genome-wide significance is not about "picking winners" or "ranking" the losers. fharrell.com/post/badb/
Image
Apr 20 • 9 tweets • 5 min read
I’ve seen critiques of the poor methodology and cherry-picking in The Bell Curve but I haven’t seen much about the absolutely deranged fever dream of predictions about the coming decades in its closing chapters. It has been 30 years, so let's review. 🧵: Image Low skill labor will become worthless, attempts to increase the minimum wage will backfire. In the not-too-distant future, people with low IQ will be a ”net drag” on society. Image
Mar 29 • 13 tweets • 4 min read
Unpopular opinion (just look at the QT's) but nearly every "dogmatic, outdated, and misleading" claim about IQ listed here is either objectively accurate or heavily debated dispute within the field itself.

Let's take them one at a time: "IQ tests were necessarily biased"

One way test bias is evaluated within the field is by testing for strong measurement invariance (i.e. that subtest behavior is consistent across groups). This method is almost never applied in the classic literature or applied poorly (MCV).
Mar 1 • 16 tweets • 6 min read
Some thoughts on the ability to distinguish populations with genetic variation, why that means little for trait differences, and why there are other good reasons to collect diverse data. 🧵 I was pleasantly surprised to see no one mount a strong defense of "biological race" in this thread. Even the people throwing this term around seem to realize it's not supported by data. Instead the conversation shifts to population "distinguishability".

Feb 27 • 5 tweets • 3 min read
Something I don't want to get lost is that the field is much better now at studying, visualizing, and discussing complex populations than it has ever been, and there are many resources to help do this effectively. A few suggestions below: The NAES report and interactive on using population descriptors [] and Coop on genetic similarity [].

Carlson et al. [] and Lewis et al. [] on accurate presentation of ancestry.nap.nationalacademies.org/resource/26902…
arxiv.org/abs/2207.11595
nature.com/articles/d4158…
pubmed.ncbi.nlm.nih.gov/35420968/
Feb 21 • 20 tweets • 9 min read
I've written about race, genetic ancestry, analyses of large biobanks, and human history



I'll summarize the key points here 🧵: gusevlab.org/projects/hsq/#…
Image Let's define some terms. Race is a social categorization of people into groups, typically based on physical attributes. Genetic ancestry is a quantification of genetic similarity to a reference population. While correlated, they have fundamentally different causes & consequences. Image
Feb 2 • 16 tweets • 8 min read
I’ve seen quotes from David Reich’s “Who We Are and How We Got Here” passed around with the insinuation that it is secretly supportive of racist and hereditarian theories, even though it directly criticizes such views. It's worth looking at what Reich actually wrote: 🧵 Reich writes at length about Nick Wade's book 'A Troublesome Inheritance', a distillation of the hereditarian position. He makes clear that Wade misleads "naive readers" into a position that has "no merit": that genetic differences correspond to traditional racial stereotypes. Image
Jan 29 • 11 tweets • 4 min read
So this is pretty typical of the low-information content you get from the genetic racists. The majority of this post is just blather but there is one (1) specific claim about genetics: that the molecular genetic contribution to IQ keeps going up every year. This is false. A 🧵:
Image The first study in 2011 into the heritability of IQ using molecular genetic methods found moderately high estimates 40-51%. But this approach was flawed technically (estimator bounds and population structure) and conceptually (environmental confounding). Image
Jan 29 • 8 tweets • 4 min read
The racists in Stancil's replies have started appealing to "scientific consensus". So let's look at what the consensus of *high-quality evidence* is on genetic racism. A 🧵: On genetics/race/behavior, over a hundred population geneticists denounced Nick Wade's A Troublesome Inheritance (a sort of genetic racism catechism). Their conclusion: "there is no support from the field of population genetics for Wade’s conjectures"

cehg.stanford.edu/sites/g/files/…
Jan 27 • 10 tweets • 6 min read
Let me expand on this since I think it's a useful lens through which think about heritability estimates. When we talk about "dominance" we're really talking about genetic effects that deviate from additivity: an effect only kicks in when you have both/neither allele. A 🧵:
Image Most common traits in humans are driven by tens/hundreds of thousands of genetic variants of small effect, so we are interested in dominance heritability i.e. the contribution of *all* of these non-additive effects together, which we can contrast with the additive contribution.
Jan 27 • 5 tweets • 3 min read
It's the year 2024 and people are still publishing twin studies with massive dominance heritability estimates -- completely implausible and not observed with any other method -- and zero data availability. Why are we still doing this?
Image Limitations: someone looked at environmental assumption violations 24 years ago so we never have to think about it again. Also, we won't even mention AC interactions, and we'll cite Purcell 2002 in a weird way so no one can find it. What do you call this act? Behavioral Genetics!

Image
Image
Image
Jan 26 • 18 tweets • 10 min read
I've written up a "crash course" on population genetics parameters useful for thinking about recent selection, heritability, and group differences (as part of a longer write-up on these concepts).



I'll summarize the key points here 🧵: gusevlab.org/projects/hsq/#…
Image A preface: if you're generally interested in population genetics it's better to learn from first principles, and I've linked some useful resources to that end (many free). In particular (spoiler) recent evolution excludes some of the more interesting concepts and personalities. Image
Jan 18 • 10 tweets • 4 min read
We discussed Duffy et al. [] in journal club. Neat approach integrating multiple sources of human genetic evidence to prioritize potential drug targets. Some thoughts 🧵:nature.com/articles/s4158… The basic idea: approved drugs are enriched for targets with multiple lines of genetic evidence: clinical, rare coding, and common GWAS. Let's put them together. (See also: Sadler, ; Minikel, ; Nelson, ; etc). pubmed.ncbi.nlm.nih.gov/37492104/
medrxiv.org/content/10.110…
pubmed.ncbi.nlm.nih.gov/26121088/
Image
Jan 12 • 17 tweets • 7 min read
Two new genetic studies of (environmentally confounded) behavioral phenotypes: Income (yes) and Educational Attainment in East Asian populations. What did they find? A 🧵:

[]
[]biorxiv.org/content/10.110…
nature.com/articles/s4156… As I've argued before () these are valid phenotypes to study but the emphasis needs to be on estimating and interpreting causal parameters, not environmental correlations. That means focusing on within-family results. So let's do that for income first.gusevlab.org/projects/hsq/#…
Jan 4 • 11 tweets • 5 min read
An academic discussion that really bugs me on here is about "administrative bloat". Which is in stark contrast to most faculty saying that they're drowning in administrative responsibilities. The problem is most people don't know how university revenue and spending works .. But there's a useful report on trends in college spending that paints a clearer picture and I recommend people engaged in this debate actually read it. A few takeaways below:

files.eric.ed.gov/fulltext/ED568…
Dec 23, 2023 • 14 tweets • 5 min read
I was curious about this idea of using genomic predictions (Enformer) as priors for eQTL discovery. So here's a quick & dirty look at that in the GEUVADIS data from Huang et al. TLDR: slightly better than promoter SNPs, much worse than Predixcan/external QTLs.

Details in thread:
Image The basic idea is we want to give more a priori weight to variants with high predictions while still controlling the Type I error. Some challenges: (1) Enformer predicts the *causal* var effect, but QTLs test marginal effects (which will aggregate multiple causal vars in LD), ..
Dec 19, 2023 • 15 tweets • 6 min read
A few thoughts on the recent set of papers torture testing genomic deep learning for predicting individual-level gene expression [ , , ]. First a brief summary 🧵:

pubmed.ncbi.nlm.nih.gov/38036790/
pubmed.ncbi.nlm.nih.gov/38036778/
pubmed.ncbi.nlm.nih.gov/38036789/
[Huang et al] evaluated four state of the art models for predicting gene expression in blood (GEUVADIS) using up to 200kb of local genomic sequence around the TSS. Using genes with known eQTLs (which were an eventual point of comparison). Image
Dec 4, 2023 • 19 tweets • 4 min read
Enjoying re-watching Day 1 of this NHGRI meeting on genetic architecture []. My notes below, with comments in []. genome.gov/event-calendar…
# Session 1:

🎤 Shamil S: Goals for the session are to discuss improving our understanding of:
1. Regulatory function
2. Low dimensional units ("pathways", "systems")
3. Evo bio: why are low fitness phenotypes still common? Image
Oct 25, 2023 • 11 tweets • 4 min read
An interesting study of social/genetic assortative mating that has implications for previous wealth persistence analyses but also highlights some subtle issues with PGI confounding. A short 🧵:

The idea is that assortative mating on a social phenotype (e.g. educational attainment), will produce correlations between mates for the causes and correlates of that social phenotype. So environment or genetics (PGI) correlated with EA will be genetically correlated in mates. Image
Oct 15, 2023 • 5 tweets • 1 min read
Admixture analysis -- where a trait value is compared to ancestry proportions -- cannot distinguish genetic from non-genetic causes. In the simulation below, two populations start with different levels of wealth, mix with assortative mating, and pass wealth down to kids. The top row plots the relationship between % admixture and the trait value, using mean/SD values from the racial wealth gap in the US. The bottom row plots the same but for a second trait, for which 20% of variance is explained by wealth and the rest is random. Image