Polygenic trait theory

Background: The theory of polygenic traits posits that many complex phenotypes, such as intelligence, height, disease susceptibility, and behavioral characteristics, arise from the additive or interactive effects of numerous genetic variants, typically thousands of single nucleotide polymorphisms (SNPs) or other loci, each contributing small incremental influences, rather than being solely controlled or solely determined by a single gene as in monogenic traits. Scholars in genetics, bioinformatics, and quantitative biology adhere to this framework to model the heritability and genetic architecture of such traits, emphasizing that no single locus dominates, which contrasts with simpler Mendelian inheritance. It is primarily used for estimating genetic risk through polygenic risk scores (PRS), predicting trait distributions in populations, guiding ?embryo selection in reproductive genetics, and informing evolutionary or epidemiological studies by partitioning phenotypic variance into genetic and environmental components. Polygenic traits are studied via genome-wide association studies (GWAS), which scan large cohorts for statistical associations between SNPs and phenotype traits using linkage disequilibrium and regression models, followed by functional validation through expression quantitative trait loci (eQTL) analysis or animal models. Polygenic trait theory does not make predictions about out-of-distribution genetic interventions.

What follows is a rant about how polygenic trait theory has been overblown, misused, and misinterpreted by both the general public and establishment scientists.

Genetic intelligence is not solely a polygenic trait. While it is true that you can do ?embryo selection based on ?polygenic risk scoring, it is also true that individual mutations or even entire genes when inserted into mammals (such as mice or rats) causes significant improvement in cognitive phenotypes (intelligence, learning, memory, etc) on various psychometric tasks. So it is not correct to say that intelligence is solely a polygenic trait. That would be like claiming that height or animal size is solely a polygenic trait, and yet we have evidence of monogenic changes that cause substantially larger and smaller litter mates to be produced.

Intelligence enhancement must be approached in a non-dumb way (x).

Polygenic trait theory comes from genetics/bioinformatics, not from genetic engineering. Religious adherence to the Church of Polygenic Trait Control is deeply dumb. Polygenic traits comes from genetics and bioinformatics. It does not come from the field of genetic engineering.

The "polygenic-only" thesis insists that, because most complex traits are controlled by thousands of loci, phenotypes can be altered only by editing a comparable number of sites; yet classic work from the 1970s-1990s shows that single-gene or few-gene manipulations routinely produce dramatic, heritable change. Cohen and Boyer's 1973 plasmid swap gave E. coli tetracycline resistance with one operon; Genentech's 1978 insulin cassette made bacteria crank out a human hormone; pronuclear injection of a single rat growth-hormone transgene doubled mouse size in 1982; Ti-plasmid delivery of one herbicide-resistance gene created the first transgenic plants in 1983; precise knock-out of individual loci in yeast and mice (Capecchi, Evans, Smithies, 1980s) linked single genes to immunity, coat color, and more; and two added carotenoid-biosynthesis genes turned white rice golden by 2000. These successes hinge on regulatory leverage, a master regulator or pathway enzyme can cascade through thousands of downstream genes, plus selectable markers that amplify rare edits and metabolic “patching” where a mini-pathway (often just 1-3 genes) yields a wholly new output. Together they debunk the notion that thousands of edits are prerequisite: targeted changes at one or a handful of loci, chosen for their outsized regulatory or biochemical impact, are sufficient to rewire cell or organismal traits. (x)

Also, the focus on polygenic traits is a total cope and a waste of time in modern academia. Just ignore them and move on. Yes, it's true that height is polygenic, but that doesn't mean you can't construct genetic circuits that increase height. (x)

the polygenic basis of genetic traits has been overhyped. basic genetic circuits can accomplish a lot! (x)

While intelligence might be a polygenic trait, it doesn't matter that it's polygenic. There are so many other interventions in genetic engineering that do not require you to use polygenic theory. (x)

We already have single edits possible to increase human cognitive performance. You don't need polygenic theory. (x)

polygenic editing is overrated and not necessary for germline genetic engineering. (x)

There's too much focus on polygenic. Just use normal genetic engineering. (x)

the interest in polygenics is overblown. Nobody cares if unmodified height is polygenic as long as modified height isn't. (x)

Why doesn't WADA do polygenic testing if they really believe in leveling the playing field? (2022-09-10)

I think finding and editing the 500 polygenic loci related to IQ (or height, etc) is a red herring. It could work, but it's like trying to jumpstart an airline industry by ducktaping a thousand birds together. (2019-02-12)

There's actually no proof that intelligence can only be modified by adherence to the polygenic theory. (x)

Height may be explainable as polygenic trait, but embryonic modification can more directly vary ?height without changing 1000s of points. (x)

Height can be achieved without modifying any of the polygenic allele sites. (x)

But actually this doesn't matter...

It turns out that even if you can modify intelligence or height without adherence to polygenic theory, it doesn't matter in the long-term because of techniques such as (1) multiplex genome editing and (2) whole genome synthesis.

(1) Multiplex genome editing refers to techniques that enable simultaneous targeted modifications at multiple genomic loci, primarily using systems like CRISPR-Cas (but also other gene editing systems), where arrays of single-guide RNAs (sgRNAs) direct the Cas9 endonuclease to specific DNA sequences for cleavage and repair via non-homologous end joining or homology-directed repair. This is achieved by delivering a multiplexed sgRNA library, often synthesized as a pooled array on a single plasmid or via viral vectors, allowing parallel editing of disparate sites without sequential interventions. For polygenic traits, this scales to hundreds or thousands of loci by optimizing sgRNA design for minimal off-target effects and using high-throughput selection or sequencing for validation. For instance, editing 1,000 loci might involve synthesizing a library of 10³ to 10⁴ sgRNAs at a cost of $10,000 to $100,000, far exceeding the $1,000 to $10,000 for simple overexpression of a single gene (e.g., via lentiviral transduction of a promoter-driven cassette, or germline genetic engineering), due to added expenses in library construction, delivery efficiency, and phenotypic screening.

(2) Whole genome synthesis involves the de novo enzymatic or chemical assembly of an entire organismal genome from oligonucleotides, typically through iterative hierarchical ligation or recombination in host cells like yeast (e.g., via TAR or Gibson assembly), as demonstrated in synthetic bacterial genomes by the J. Craig Venter Institute. DNA fragments (50 - 100 bp oligos) are synthesized commercially using ?DNA synthesis techniques, assembled into larger scaffolds (1 to 10 kb), then chromosomes (Mb-scale), and transplanted into recipient cells for replication and expression, enabling arbitrary redesign of all genetic elements. This allows comprehensive editing of every locus by rewriting the genome, but at prohibitive costs of (presently) $0.01 to $0.10 per base pair (totaling hundreds of millions of dollars for the 3 to 6 Gb human genome), versus $100,000 to $1 million for multiplex editing of thousands of loci, making it economically viable only for microbial or small viral genomes today while simple interventions like single-gene knock-ins remain orders of magnitude cheaper.

Depending on which techniques or approaches you are using, it may not be economical to use polygenic theory. If you are only able to do ?embryo selection then polygenic trait theory becomes a lot more valuable. If you are able to use modern molecular biology techniques, then it is cheaper to use genetic engineering techniques.

Caveats

While it is true that genetic engineering can cause changes through monogenic changes, single nucleotide polymorphisms, single point mutations, small deletions, large deletions, copy number variations, transgenic overexpression, knock-in, knock-out, silencing, repressors, etc... it's also true that polygenic traits exist. It depends on what your goal is. If your goal is to study polygenic traits, then you don't need genetic engineering at all. If your goal is to modify intelligence or height, then the opposite applies and you may not need that much knowledge of the polygenic basis of these traits.

There is still pleitropy, though. Pleiotropy is a genetic phenomenon in which a single gene or mutation influences multiple, often seemingly unrelated phenotypic traits, with implications for understanding complex diseases and evolutionary biology in molecular genetics and bioinformatics. And you still need to do testing of some kind. Quality control still matters. Being right still matters. Empiricism is important. Finding a strong SNP is kind of unlikely, but sometimes it does happen. Often you will need to move to a whole genetic circuit, or introducing a new gene, or modify the protein via protein engineering methods. Stuff like that. When you are looking at a mutation from a GWAS study, keep in mind that the study might be poorly designed, underpowered, or the association may not be causative, etc. The mechanistic basis of action still matters.

There are further caveats listed on the germline genetic engineering proposals page.