Example of eQTL statistics.

lgBFx columns are log10 BFx. Column βj’s are effect estimates in joint analysis used to compute BFj. P column contains − log10 p-value with the null hypothesis β0 = β1 and alternative hypothesis β0 ≠ β1. Transcript start site (TSS) is in Mb, and the coordinate is from HG38. Bona fide imprinted genes are highlighted.

Comparison of paternal effects and maternal effects of sentinel eQTL in joint analysis.

Each point is the most prominent eQTL with log10 BFj > 3 of gene. The paternal effect is on x-axis and maternal effect y-axis. The coloring reflects significance of differentials between paternal effects and maternal effects, with red more significant than blue.

An example of gene that whose paternal and maternal eQTL have opposite effect sizes.

On the left is paternal vs maternal effect sizes normalized by their corresponding standard deviation. The effect sizes were estimated by joint analysis. The non-significant eQTL are colored in gray and significant eQTL colored in black. An eQTL marked in blue was chosen to show boxplots with gene expression (phenotype) of NECAB3. The three panels of boxplots are for genotypes, paternal alleles, and maternal alleles. For this SNP rs4911348, the genotypes has no association with phenotype, but both paternal alleles and maternal alleles are associated with the phenotype, and the paternal and maternal effects are in opposite direction. Right panel bottom show Manhattan plot of cis-eQTL of the gene NECAB3. The genotype Bayes factor were colored in gray and joint test Bayes factor were colored in black. SNP rs4911348 was highlighted in blue.

Top tabular: Annotation of eQTL. SO is a set of opposing eQTL, SP is a set of paternal eQTL, SM is a set of maternal eQTL, and SG is a set of genotype eQTL. The column eQTL contain counts of eQTL in each set, and column SNP contains counts of distinct SNPs of eQTL in the set. The column eGene is the number of genes associated with eQTL in the set. The column of Len is the median length in Kb of the eGenes. (The pattern is the same with the mean length.) The column DTSS contains median distance in Kb to transcription start site. The column GWAS contains percent of GWAS hits among SNPs, with GWAS p-value threshold of 5 × 10−8. Bottom tabular: A SNP set Pk contain SNPs in SG that are eQTL of at least k eGenes. (Note P1 = SG.) Percent of GWAS hits for SNP set Pk increases with k.

Enrichment of drug target genes.

GO contains eQTL with both paternal and maternal effects but they are in opposite direction. GP contains eQTL of paternal effects. GM contains eQTL of maternal effects. GG contain eQTL that have both paternal and maternal effects and they are in the same direction. Column nGene are sizes of gene sets, Column nTarget are counts of successful drug targets. The last column contains one-sided test-of-proportion p-values. The highlighted p-values are significant after Bonferroni correction of 12 tests.

Statistics and accuracy of phasing triple heterozygous sites via mask and imputation.

Ni: number of simulated samples. Ns number of biallelic SNPs used in simulation. N3H : number of triple heterozygous SNPs across Ni samples. NT : number of triple heterozygous SNPs that being correctly phased. R: ratio of the correctly phased triple heterozygous SNPs. EUR: European samples. EAS: East Asian samples. AFR: African samples.

Examples of eGenes harboring different sets of eQTL: GZMH and DSE.

Each gene has two plots: a square plot showing normalized paternal effect (x-axis) vs normalized maternal effect (y-axis), and a rectangle plot showing test statistics along chromosome position. In each plot, gray dots are insignificant eQTL, black dots are significant genotype eQTL, blue dots are significant paternal eQTL, and green dots are significant maternal eQTL. The vertical line in the right panels mark the transcription start site.

Kinship and sd.

The x-axis is twice of kinship averaged over 22 estimates, one for each autosome. The y-axis is standard deviation of those 22 estimates. Plot only show relevant portion of the kinship.

GC content bias correction.

Top panel is before correction; Bottom left is after correction with local linear regression; Bottom right is after correction with local quadratic regression (loess in R). Each line represents fitted values of one sample.

Annotation of eQTL for different threshold.

Top tabular: θ = 0. Middle tabular θ = log10 2. Bottom tabular: threshold= θ = log10 3. SO is a set of opposing eQTL, SP is a set of paternal eQTL, SM is a set of maternal eQTL, and SG is a set of genotype eQTL. The column eQTL contain counts of eQTL in each set, and column SNP contains counts of distinct SNPs of eQTL in the set. The column eGene is the number of genes associated with eQTL in the set. The column of Len is the median length in Kb of the eGenes. (The pattern is the same with the mean length.) The column DTSS contains median distance in Kb to transcription start site. The column GWAS contains percent of GWAS hits among SNPs, with GWAS p-value threshold of 5 × 10−8.

Counts of eQTL and eGenes for different Bayes factor threshold.

As the threshold increases, type I error rate reduces, power reduces as a consequence, so the numbers of positive eQTL and eGene reduce.

Allele frequecy distribution for parental eQTL and maternal eQTL.

Here we used allele frequencies of European descent from 1000 Genomes project.