Cross-tissue atlas of ribosomal protein eQTLs

Abbreviations and definitions

Gene-Tissue Expression (GTex) Project
Ribosomal protein (RP)
expression quantitative trait loci (eQTL)
significance threshold for analyses here (Q-value for ‘eGene’ status <0.05 & nominal P<5x10-6)

Forest plots for RP eQTL effect sizes

pink = decreased expression
seagreen = increased expression

Mitochondrial RP eQTL effect sizes (large subcomponent)

Mitochondrial RP eQTL effect sizes (small subcomponent)

Cytosolic RP eQTL effect sizes (large subcomponent)

Cytosolic RP eQTL effect sizes (small subcomponent)

Decreased expression

Increased expression

Type

large or small RP in cytosol or mitochondria by direction of eQTL effect

	decrease	increase
MRPL	480	312
MRPS	156	204
RPL	343	84
RPS	552	158

Mosaic plots

\[X^2=\sum \frac{(o-e)^2}{e}\]

Based on chi-squared tests
The two mosaic plots display the same information, but emphasize the relationships differently.
Blue color indicates that the observed value is higher than expected
Red color specifies that the observed value is lower than expected

Same information, collapsed (mitochondrial vs cytosolic)

Mosaic for “bivalence” and direction

	no	yes
MRPL	288	504
MRPS	118	242
RPL	308	119
RPS	336	374

	decrease	increase
no	760	290
yes	771	468

“Bivalence” (RP gene having both decreasing and increasing expression eQTLs)
These plots demonstrate that RP genes with without bivalent eQTLs are more likely than chance to decrease expression and less likely than chance to increase expression.

Mosaic for EAF

These plots show that for EAF <0.50, an eQTL is more likely than expected to decrease expression; for EAF>=0.50, an eQTL is more likely to increase expression.

But we could subset the data annd look at what’s happening by EAF <0.05 (vs 0.50). . . .

Deleteriousness

Combined Annotation Dependent Depletion (CADD) provides the predicted functional impact of noncoding variants.

The PHRED-like (-10*log10(rank/total)) scaled C-score ranks a variant relative to all possible substitutions of the human genome (8.6x10^9). A score of >=10 indicates that these are predicted to be the 10% most deleterious substitutions that you can do to the human genome and a score of >=20 indicates the 1% most deleterious.

	bottom	top 10%
cytosol	819	139
mitochondra	929	55

Zooming in on a particular gene

I have highlighted by SNP.

Gene with both increased & decreased eQTLs

Overall effect sizes with the -log10(P)-values

To make the effect size axis center at zero, I created a dummy row and later erased the observation it generated.