W9L2 Thu Post-omnigenic consideration Flashcards
What have we observed from human population genetic / evolution
- Humans genome have been very much shaped by demography and natural selection.
- There is lots of rare variation segregating within living humans, and we are still learning to predict its phenotypic consequences.
What have we observe from GWAS
- Complex traits are complex (many gene involved).
- Most loci contribute small effects to quantitative traits.
What have we observe from functional genomics
- Gene regulatory processes are fundamental to establishing cell and tissue identity.
- Some variation in gene expression can be linked to genetic differences between individuals.
The omnigenic model of human traits
- Complex traits are the product of core and peripheral effects spreading through a network
- Any given peripheral gene will contribute little, compared to a given core gene… But there’s many of them!
* So as a whole, peripheral genes contribute a lot! - Since all genes are connected, most genes contribute to most traits
* More nuance: expressed genes contribute to traits in cell-type specific ways
The omnigenic model problem? human height example
- Looking at GIANT data, 62% of all common SNPs are associated with non-zero effect on human height.
- But if all SNPs really do contribute to everything, why/how do different traits exist
The architecture of human height
- The more SNPs a SNP is in LD with, the more likely it is to appear to contribute to height.
- A better interpretation would be 62% of genomic windows (10-100kb wide) contribute to height
- After some math (Boyle et al supplement), predict ≈100,000 SNPs genome-wide contribute to height.
Finding disease causal SNPs
- Consider all GWAS SNPs for a specific disease.
- Are they located in genetic regions active only in certain cell types, or broadly?
- (Active refers to functional genomics data from Roadmap Epigenomics)
- And what cell types?
Defining core and peripheral genes, past deifinition
- we might think of core genes as the genes that (if mutated or deleted) have the strongest effects.
- Or we might think of core genes simply as the genes with interpretable mechanistic links to disease
How predictive are eQTLs
-mRNA expression is only the first step on the way to steady-state protein levels
* Variation at early stages can be buffered by downstream processes.
* Don’t have complete info to say how much, or how often
* Genetically-driven differences can also be decoupled from RNA expression
Defining core gene, current
We define a gene as a ‘core gene’’ if and only if the gene product (protein, or RNA for a noncoding gene) has a direct effect
—not mediated through regulation of another gene
-affect cellular and organismal processes leading to a change in the expected value of a particular phenotype
eQTLs vs human evolution
- Most genes have eQTLs because they can afford to have eQTLs, evolutionarily speaking. * Most changes in expression are, individually, neutral
Why do we even map eQTL
- eQTLs that have large effect sizes, or that are active across multiple tissues, are unlikely to be involved in disease aetiology.
- eQTLs with more restricted effects are still likely to be interesting.
exp: SNPs lead to difference in expression under stimulus
eQTL in early development
-Look at the eQTL at different stage of development
Why do traits have a complex architechture
-many trait are continuous, unlikely to be a simple traits
-peripheral snps change would not affect the final phenotype much-> does not affect optiumum
-if a trait is control by a few core gene only, SNPs can lead to loss of optimum after a few generation
-This is backup by the fact that many snps dose not have negative effect, only some have negative selection
