06 Polygenic Disease Flashcards
Name some polygenic diseases
Diabetes, heart attacks, depression
Continuous characteristics are those that are ____________________ determined by genetics
partly
What was the polygenic model made by RA Fisher in 1918?
The idea that many mendelian genes each with a small effect individually would produce the continuous characteristics
Even dichotomous characteristics like whether I have diabetes or not is dependent on what?
Many genes, as many genes contribute to susceptibility
Define heritability
How far variations in a characteristic are due to genetic differences, and how far they are due to environmental factors. A number between 0 and 1.
Why is heritability a difficult, complex concept?
Because the social conditions are important. In a country where there are large inequalities, you would say that the environmental factors have a higher contribution. A trait would seem more heritable in a more equal society.
Family studies are one way to estimate heritability, because we share 50% of our DNA, but we also share an environment! What is lamdaS?
Your risk score. The risk of you having something with an affected sibling compared to the risk of the general population.
What is the risk score of breast cancer, CF, and huntingtons?
Breast cancer - 2,
Cystic Fibrosis - 500
Huntingtons - 5000
Why are twin studies better than family studies?
Because you share even more DNA if monozygotic, and probably share a more similar upbringing than other siblings
How can you eliminate some biases in twin studies?
Study only same sex twins if two zygotes. Be aware that some DZ twins might be classified as MZ twins incorrectly.
What are the two possible study designs of Adoption studies?
1 - Find children of affected parents who were adopted at birth. Did they ‘escape’ the family disease.
2 - Find children who have the condition but were adopted at birth. Does the condition run more in the birth family or the adoptive family?
What’s a general way to identify genetic susceptibility factors?
Get 10,000 unrelated cases and 10,000 unaffected controls. Age and gender match them. Type them all for a large number of SNPs. Look for variants associated with disease
1 in how many nucleotides vary frequently in the population?
1 in 300. The other 299 have rarer variants.
Why didn’t initial studies into susceptibility factors find much that could be replicated?
The P value was too large.
What study made a big impact into identifying susceptibility factors
WTCCC in Nature 2007. Looked at 14,000 cases of 7 common disorders (Bipolar, CAD, Crohn’s disease, hypertension, rheumatoid arthritis and Type 1 and type 2 diabetes). They looked at 500,000 SNPs with a P value of 5x10^-8.
What is the odds ratio for SNP association studies?
Ratio of being a case rather than a control with the variant compared to odds of being a case rather than a control without the variant.
A measure of how much the variant contributes to whether you have the disease or not.
What did the WTCCC find from their study of 14,000 individuals with 7 common diseases?
24 regions with genetic associations with disease, most were in non-coding regions.
How do you plot the results of a GWAS?
A manhattan plot. X axis is the length of the chromosome. Y value is the -log p value.
GWAS needs you to be very stringent with what?
Diagnosing who is a case and who is a control. Having a large sample size. Having a small p value cut off. Quality controls of genotypes.
True or false, nearly every disease has been studied by GWAS?
True, just look at the EBI GWAS catalog
How are we increasing GWAS power?
People are pooling lots of data together as we test more and more people, so we can identify weaker associations and rarer variants.
What have we learned about where most variants are from GWAS studies?
They are in non-coding sequences, probably regulatory sequences. This means we aren’t always sure what genes they are affecting.
When you identify a gene in GWAS. What can this give insight into?
The biology of the condition
Early studies trying to use susceptibility factors were not useful in predicting people’s risk, what is there a shift to do now?
Use every variant in the genome, even those we have no idea what most of them might have to do with the disease.
How are polygenic risk scores made? (Broad kicked this all off)
Use a big biobank with population scale data of affected and unaffected people, and get a computer to devise the best estimators of disease based on the genetic data to then distinguish between other people.
What did the big Broad study of polygenic risk scores look at? What diseases, and how many people were involved?
Looked at CAD, Atrial fibrillation, Type 2 Diabetes, Inflammatory bowel disease, breast cancer.
Created great model from 100,000 people, then tested against 300,000 people.
