Whole genome sequence association and ancestry-informed polygenic profile of EEG alpha in a Native American population

Abstract

EEG alpha activity is the dominant oscillation in most adult humans, is highly heritable, and has been associated with a number of cognitive functions. Two EEG phenotypes, low- and high-voltage alpha (LVA & HVA), have been demonstrated to have high heritabilities. They have different prevalence depending on a population's ancestral origins. In the present study we assessed the influence of ancestry admixture on EEG alpha power, and conducted a whole genome sequencing association analysis and an ancestry-informed polygenic study on those phenotypes in a Native American (NA) population that has a high prevalence of LVA. Seven common variants, in LD with each other upstream from gene ASIC2, reached genome-wide significance (p = 2 × 10^-8 ) having a positive association with alpha voltage. They had lower minor allele frequencies in the NAs than in a global population sample. Overall correlations between lower degrees of NA (higher degree European) ancestry and HVA, and higher degrees of NA and LVA were also found. Additionally a rare-variant gene-based study identified gene TIA1 being negatively associated with LVA. Approximately 3% of SNPs exhibited a 15-fold enrichment that explained nearly half of the total SNP-heritability for EEG alpha. These regions showed the most significant anti-correlations between NA ancestry and alpha voltage, and were enriched for genes and pathways mediating cognitive functions. Our findings suggested that these regions likely harbor causal variants for HVA, and lacking of such variants could explain the high prevalence of LVA in this NA population, possibly illuminating the ancestral origin and genetic basis for EEG alpha.

Keywords: admixture; gene based rare variant analysis; genome-wide association; heritability enrichment; low coverage sequencing.

FIG 1

Ancestry distribution by variant genotypes. The width of the boxes is proportional to the squared-root of the number of individuals in each group. The genotypes (reference/alternative allele) in each panel are A: G/A, B: C/T, C: G/C, D: G/T, respectively. The variants in panels A, B, and D are positively associated with high-voltage alpha. Variant rs78245818 in panel C is positively associated with low-voltage alpha.

FIG 2

EEG alpha voltage is anti-correlated with the degree of Native American ancestry. Left panel: alpha voltage vs. NA, r = −0.119 (p = 0.0017). If alpha voltage is residualized over age, age-squared and gender, r = −0.107 (p = 0.0049). Right panel: alpha voltage was winsorized, residualized over age, age-squared and gender, and standardized, r = −0.132 (p = 0.00049). The fitted lines are from the linear regressions.

FIG 3

The estimated degrees of local Native American ancestry for 4396 overlapping windows along the chromosomes. Horizontal lines correspond to mean (μ), mean ± 1, or 2, or 3 standard deviations (σ) of the correlation coefficients, respectively.

FIG 4

Pearson’s correlation coefficients between EEG alpha voltage and the local Native American ancestry, for 4396 overlapping windows along the chromosomes. Horizontal lines correspond to p values at 0.03, 0.01, 0.001, and 0.0001, respectively.

FIG 5

Joint heritability estimates with genomic partitions. A. Joint heritability of genome partitions. The first bar on the left is the estimated total SNP-heritability of the whole genome. B. Joint heritability of genome partitions and thresholded genetic covariance matrix (genetic relation > 0.05). The left bar in each group is the heritability estimate on the thresholded genetic covariance matrix. The second bar in the first group of bars on the left is presumably the estimated total SNP-heritability when controlling for relatedness. The remaining six groups of bars in each panel correspond to the heritability estimates of the six sets of 2-partitions in the same order detailed in Table III.

FIG 6

P-values of logistic regressions of the low-voltage alpha, Low10, and the high-voltage alpha, High50, onto the degree of local NA ancestry for the 78 windows that showed most significant negative correlation with p ≤ 0.0001 between alpha power and local NA ancestry.