BSMS205 · Genetics · Final Lecture

QTLs

Chapter 30 · Part V · Functional Genetics

Today's central question

If you carry a variant,
how does it actually change
RNA or protein levels?

What QTLs are

Quantitative Trait Locus = variant linked to a measurable trait
eQTL · variant ↔ mRNA level
pQTL · variant ↔ protein level
The bridge from genotype to molecular phenotype

Roadmap for the final lecture

How QTL mapping works · cis vs trans
eQTLs and pQTLs · what each tells you
Three landmark studies from 2025
Caveats · what QTLs do not tell you
Integrating with GWAS · finding causal genes
The complete genetics framework

§ 1

How QTL Mapping Works

Three ingredients

Genotype data · millions of SNPs per person
Molecular phenotype data · RNA-seq for eQTL · mass spec for pQTL
Statistical test · linear regression of trait on genotype

The additive model

Genotype	Encoding	Expected effect
AA	0	baseline
AG	1	baseline + β₁
GG	2	baseline + 2β₁

Cis-QTL vs trans-QTL

Cis

Variant near the gene
Within ±1 Mb
Strong, easy to detect
Usually a regulatory element

Trans

Variant far away · different chromosome
Weaker effects
Often via a TF or signaling intermediate
Harder to detect

Multiple testing

Millions of SNPs × thousands of genes = billions of tests
Most associations are noise
Use FDR or permutation thresholds
Large samples needed for trans-QTLs (~50,000+)

§ 2

eQTLs and pQTLs

eQTLs · genetic effects on RNA

Most disease GWAS variants are actually eQTLs
Don't change protein sequence · change expression level
Tissue-specific · GTEx mapped 54 tissues

GTEx Consortium 2020, Science

pQTLs · genetic effects on protein

Measured in plasma by mass spectrometry
Protein is closer to phenotype than RNA
Captures translation efficiency, stability, secretion
Only 40 – 60% of eQTLs have a matching pQTL

Why eQTL and pQTL diverge

Translation efficiency · some mRNAs translate more
Protein half-life · some proteins degrade fast
Post-translational modifications · alter activity, stability, location
Secretion · cellular vs plasma protein

§ 3

Three Landmark Studies

Study 1 · Niu et al. 2025 · childhood pQTLs

2,147 children and adolescents · plasma proteomics
1,216 high-confidence proteins measured
~70% of proteins influenced by genetics, age, sex, or BMI
1/3 of proteins have at least one significant pQTL

Niu et al. 2025, Nature Genetics

Niu 2025 · genome-wide pQTL map

Niu 2025 plasma pQTL Manhattan plot — Most pQTLs are **cis** · in non-coding regulatory regions · only ~3% are missense.
Niu et al. 2025, *Nat Genet*. CC BY 4.0.

Study 2 · Hofmeister et al. 2025 · parent-of-origin

UK Biobank · ~109,000 participants
14,000 known pQTLs tested for parent-specific effects
30+ significant POEs identified
40% of POEs are antagonistic · maternal vs paternal push opposite directions

Hofmeister et al. 2025, Nature

Second study, Hofmeister and colleagues twenty twenty-five in Nature. They tested whether p Q T Ls show parent-of-origin effects — cases where the maternal and paternal alleles have different effects. Standard genetic models assume the parent does not matter. But for imprinted genes, only one parental copy is expressed. Using one hundred nine thousand U K Biobank participants, they tested fourteen thousand known p Q T Ls and identified more than thirty significant parent-of-origin effects. Forty percent showed antagonistic effects — maternal and paternal alleles pushing the trait in opposite directions. Classical imprinted loci like H one nine I G F two, K C N Q one, and K L F fourteen M E S T were all confirmed. The findings replicated in the Estonian Biobank and the Norwegian Mo Ba cohort.

Study 3 · Wingo et al. 2025 · multi-ancestry brain pQTLs

1,362 brain donors · African-American, Hispanic, non-Hispanic White
~11,750 proteins quantified
858 fine-mapped causal pQTLs
119 pQTL-protein-trait triads link variants → protein → disease

Wingo et al. 2025, Nature Genetics

Wingo 2025 · ancestry-specific pQTLs

Wingo 2025 multi-ancestry brain pQTLs — Some pQTLs are **ancestry-specific** · some are **shared**.
Wingo et al. 2025, *Nat Genet*. CC BY 4.0.

§ 4

Caveats

Caveat 1 · association is not causation

QTL = correlation between variant and trait
True causal variant may be in linkage disequilibrium
Or the variant may act via an intermediate
Need fine-mapping and functional validation

Caveat 2 · tissue and cell-type specificity

QTLs are often context-dependent
APOE eQTL active in liver · not the same elsewhere
IL2RA eQTL only after cytokine stimulation
Always check the right tissue · the right context

Caveat 3 · effect sizes are small

Cis-eQTL: typically 5 – 20% of expression variance
Trans-QTL: <1%
pQTL: usually 1 – 5%
Small effects · but they accumulate across thousands of variants

§ 5

QTL × GWAS ·
Finding Causal Genes

The integration workflow

GWAS finds a disease locus (e.g. chr 19, Alzheimer's)
eQTL / pQTL mapping shows the same variants affect APOE and CLU
Colocalization · GWAS and QTL signals share a causal variant
Strong evidence: APOE/CLU dysregulation mediates AD risk

Mendelian randomisation

Use genetic variants as instrumental variables.
Test whether changing a molecular trait causally affects disease risk.

If raising APOE protein reduces AD risk → APOE is a drug target

§ 6

The Complete Framework

The full chain · variant to disease

DNA variant → RNA expression (eQTL) → Protein level (pQTL)
→ Pathway activity → Cellular phenotype → Disease risk

Course recap · Part IV

Ch 20 · allele frequency — the fundamental statistic
Ch 21 · population structure — why frequencies vary
Ch 22 · linkage — Mendel to Morgan
Ch 23 · recombination and haplotypes — how variants travel
Ch 24 · VCF — how all of this becomes data

Course recap · Part V

Ch 25 · forward genetics — phenotype to gene · GWAS · burden tests
Ch 26 · reverse genetics — CRISPR toolkit · screens · rescue
Ch 27 · CRISPRa therapy — SCN2A as the case study
Ch 28 · gene regulation principles — promoters · enhancers · loops
Ch 29 · regulation methods — RNA-seq · ChIP-seq · ATAC-seq · CUT&RUN
Ch 30 · QTLs — connecting variant to molecular phenotype

What modern human genetics can now do

Find variants associated with disease at population scale
Determine which gene each variant regulates
Determine in which tissue and cell type
Test the gene's function with CRISPR
Design therapy by adjusting expression of the right gene

§ 7

Summary

What to take away

QTLs link variants to molecular phenotypes
eQTLs and pQTLs · 40 – 60% concordance
Most are cis · most non-coding · most tissue-specific
Effect sizes small · but accumulate across the genome
QTL + GWAS = causal gene identification
Mendelian randomisation → causal inference for therapy

End of course

From A and T and G and C
to human disease.

BSMS205 · Genetics · 2026 Spring · 30 chapters complete.