GADGET

The GADGET web server is designed to allow users to explore the genetic basis of human phenotypic diversity. GADGET integrates trait-implicated single nucleotide polymorphisms (SNPs) from GWAS, with population genetic data from the 1000 Genomes Project (1KGP), to calculate genome-wide polygenic trait scores (PTS) for hundreds human phenotypes in thousands of individual genomes. Users are able to (1) explore the global of distributions of pre-computed genome-wide polygenic trait scores (PTS) for a wide variety of human phenotypes, and (2) compute worldwide PTS distributions for user-defined traits of interest.

Heritability and human population genetic variation

Heritability is the amount of phenotypic variation that can be attributed to individual genetic differences; all human traits that have been measured thus far show evidence for some degree of heritability. Given that (1) the expression of heritable traits is influenced by genetic variants, and (2) the frequencies of most genetic variants vary among human populations, heritable traits are expected to differ among populations. Indeed, there are numerous examples of heritable traits that vary widely across human populations.

The genetic architecture of human traits

The genetic architecture underlying heritable human traits can be deciphered via genome-wide association studies (GWAS). Hundreds of GWAS studies have been used to discover thousands of genetic variants (SNPs) that influence the expression of a wide variety of human traits, including anthropomorphic, behavioral, and health-related phenotypes. We mine the NHGRI-EBI GWAS Catalog, the PubMed database, and the EBI Experimental Factor Ontology to curate sets of SNPs that shape the expression >800 human traits. Trait descriptors parsed from the Experimental Factor Ontology are hierarchically organized into a custom ontology, which can be explored here: Jordan Lab Trait Ontology.

Human genome variation worldwide

The 1000 Genomes Project (1KGP) provides a deep catalog of global human genetic variation. The 1KGP characterized complete genome sequences for 2,504 individuals from 26 global populations, organized into five continental (super) population groups: Admixed American (green), African (blue), East Asian (red), European (yellow), and South Asian (brown). We consider the presence of trait-implicated SNP effect alleles in individual human genomes to calculate genome-wide polygenetic trait scores (PTS), and then we evaluate the distributions of PTS across global populations.

Genome-wide polygenic trait scores (PTS)

Genome-wide polygenic trait scores (PTS) have emerged as a powerful tool for predicting individuals’ phenotypes based on the numbers of effect (risk) alleles encoded in their genomes. PTS can be computed by summing the numbers of effect alleles encoded in an individual genome, and scores can be weighted by considering allele effect sizes. In the case of health-related phenotypes, PTS are often referred to as genetic risk scores, reflecting the predicted health risk to individuals entailed by the presence of disease-implicated variants in their genomes. Unweighted PTS ($uPTS$) are calculated as the normalized sums of the number of effect alleles found in the genome for all trait-associated SNPs: $$uPTS={\sum }_{i=1}^n{EA}_i/{\sum }_{i=1}^n{A}_i$$ where ${EA}_i \in \left\{0 , 1, 2\right\}$ corresponds to homozygous absent, heterozygous or homozygous present risk alleles at each SNP and $A_i \in \left\{0 , 1, 2\right\}$ corresponds to the total number of alleles with basecalls at each SNP. Weighted PTS ($wPTS$) employ effect size estimates, either odds ratios or β-values, to weight the numbers of observed effect alleles for each trait associated SNP:
$$wPTS={\sum }_{i=1}^n{\mathrm{(E}A}_i\times {es}_i)/{\sum }_{i=1}^n{A}_i$$

where ${es}_i$ is the SNP-specific effect size estimate.