what does the spectrum of disease etiology explain?
disease etiology ranges from conditions w/ purely genetic, single causes (hemophelia, CF - single gene memdelian disorders) to totally environmental (motor vehicle accident)
multifactoral causation, when multiple genetic loci in varying contributions contribute to complex disease risk, is most challenging to study re: overall risk of disease development
difference between mendelian model and complex disease model re: "genetic" disease
mendelian: gene mutations are causal - sufficient and necessary - to cause disease
complex disease: gene muation/variation predisposes to certain disease, or increases chances; not sufficient or necessary
how can one determine heritability?
genetic phenotype / total variance = degree of genetic determination = heritability
what kind of study can determine if a phenotype of clinical interest is dependent on genetic variants' heritable component?
compare heritability in monozygotic twins vs. dizygotic twins
MZ 100% twins share genes, intrauterine environment, household
DZ twins share 50% genes, 100% intrauterine environment, household
thus can compare heritability btwn twin types
formula for heritability estimates w/ twin studies?
h2 = 2*(r2MZ - r2DZ)
heritability estimate = correlation between concordance of having disease btewn monozigotic and dizigotic twins
what is the heritability range?
0 < heritability < 1
heritability = 0 if trait is not genetic at all (offspring resemble general population more than they do the parents)
heritability = 1 if trait is totally genetic (offspring resemble parents more than they do the general population)
if you think your phenotype is heritable, how can you test that hypothesis?
candidate gene approach:
1) look for candidate gene in known pathways and then
2) knock out that gene in animals w/ similar phenotypes
3) see effect of knock out
how is insulin signaling pathway an example of the candidate gene approach?
to study diabetes, can look at genes involved w/ insulin signaling pathway, study how/if ligand binding works/is implicated
can select for candidate genes based on knowledge of pathway
how does candidate gene study via genetically altered animal model work?
knock out insulin receptor to study its impact on growth/life
see at birth, IR-deficient pups are indistinguishable from WT or heterozygous pups
after suckling, see major metabolic alternations begin
IR- pups develop severe form of diabetes, growth retardation, skeletal muscle hypotrophy, death 1 week post-birth
what is best way to test if a particular variant is sssociated w/ a disease of interest?
case-control association study testing frequency of a specific allele between genotypes at the genomic position of interest among disease cases and matched (age, ethnicity, gender) cohort of unaffected individuals
how are SNP distributions among cases'/controls' genotypes experimentally compared in case-control association study?
obtain genotypes at genomic positions of interest (e.g. polymorphic variants in a candidate gene coding region or in regulatory elements) in a pool of cases and of controls
see if there are differences in specific allele frequnecy btwn cases and controls; this would suggest an association btwn disease and allele, whereas if have similar frequency between cases and controls, no association btwn disease and allele
what does an association studie determine?
if the variant increases the risk of disease and it's associated
helps under disease's etiology
is not predictive of having disease or not based on the variant
study in which density of genetic marker is sufficient to capture a large proportion of the common vartion in the human genome in the population under study
so can genotype a large number of variants genome-wide
what does GWAS use to do study?
collects allele frequency data from common variants distributed across the entire genome in large cohorts of cases and controls for diseases of interest
computers significance scores for each variant
what are affymetrix and illumina?
platform for high throughput genotyping in a GWAS
what results does GWAS produce?
OR: ratio in differences in the frequencies of having the disease, between cases and controls; aka the effect of having the SNP
accompanied by p-value
plotted on a Manhattan Plot
describe a manhattan plot
plots computed significance scores for each variant tested in a GWAS as a function of genomic location, aka chromosome and position
each dot = the p-value for a statistical test for 1 SNP
b/c log scale, higher plot = lower p-value = more significant
horizontal line: genome-wide statistical significance; look at results above the line
what did the advanced macular degeneration GWAS look at / find?
a common coding variant, Y402H, in the complement factor H (CFH) gene on chromosome 1 (1q31) increases the risk of developing AMD
studies estimated the odds ratio associated with this variant for all categories of AMD to be between 2.45 to 3.33
the odds ratios were higher, between 3.5 and 7.4, for advanced dry and wet forms of AMD
CFH inhibits the formation and accelerates the decay of alternative pathway C3 convertases and serves as a cofactor for the factor-1 mediated cleavage and inactivation of C3b
how does sample size pose a challenge to GWAS?
need very large sample b/c have low threshold for p-values
if effect is small, = 1.2, means carrying variant increases your risk of disease by 20%
to detect such a rare variant, may need as many as 7,000 ppl in study
if effect size is large (i.e. OR=2.0), could have smaller sample size (i.e. n=1,000)
are GWAS effective at explaining common variants? why/why not?
GWAS analysis of common diseases has shown that it explains only a few percent(s) of the genetic component of disease heritability, = "missing heritability"
aka other things are contributing to heritability that the GWAS cannot show
GWAS explains only a few genetic components
what does the common disease-common variant vs. common disease-rare variant hypothesis explain?
why GWAS are only somewhat informative, and unhelpful for common diseases' explanation
hypothesis is: common variants will explain phenotypes in common diseases
actually: we cannot explain lots of heritability
very rare variants that cause disease, we know about = mendelian disease
and common variants w/ small effect size have been described by GWAS
however GWAS cannot explain low-frequency variants w/ intermediate penetrance/effect size
what are challenges of association studies?
1) require lare sample sizes (thousands of individuals) to attain significance
2) require phenotypically homogenous groups of cases & controls
3) sensitive to populations stratification
4) produce false positive results
5) hard to replicate
what's the most important predictive tool? why?
family history is risk factor for many chronic diseases
family history useful to assess health risk, initiate interventions, motivate behavioral changes
lower cost, greater acceptability, reflection of shared genetic and environmental factors make it better than genomic tools
also see that genetic tests add very little predictive value compared to using clinical risk scores based on demographic, family history, and clinical/laboratory data
challenges for next gen sequencing?
1) rare variants require larger sample sizes
2) millions of variants require multiple testing
3) lack of good annotation
4) limited statistical approaches exist to identity associations w/ phenotypes of interest
5) value comes from analysis, annotation, and association w/ other data; raw genetic data has little value
how did GWAS reposition Crohn's disease pharma work?
determined new target, IL23R gene on chromosome 1, significant in Crohn's association study
used Stelara, drug for cirrhosis, for Crohns b/c targeted interleukin 12 and interleukin 23 - gave Crohns patients who'd been untreatable improvements in symptoms w/in 6 weeks
PCSK9 / LDL
explain how PCSK9 works re: LDL
what did GWAS show
LDL taken into cell for degredation by the LDL receptor
usually, LDL receptor is recycled back to cell surface to endocytose more LDL after takes in some LDL
PCSK9 is gene that causes degredation of LDL receptors
in people who have high levels of LDL in plasma, want to thus inhibit PCKS9 activity so that LDL receptors aren't destroyed, and instead are recycled to endocytose/degrade more LDL
thus individuals w/ high cholesterol who have PCSK9 mutation have lower LDL eventually; aka could block PCSK9 in people with high cholesterol to uptake more LDL
are SNPs predictive of ype 2 diabetes risk?
number of risk alleles counted in diabetes vs non-diabetes persons
lower genetic risk score calculated, = more likely to be a control
if have very high genetic risk count score, almost definitively a carrier w/ a genetic risk
what does this show?
gene count does not predict as well for diabetes as framingham standards do
family heritability thus is most important predictor of inheritance, more than gene count score
what can modify genetic risk/effect?
gene-diet - nutrigenomics
gene-drug - pharmacogenetics
gene-lifestyle - smoking, exercise
what does FTO gene study show?
FTO: fat mass and obesity-associated protein
AA: carries 2 risk variants for obesity
GA: carry 1 risk allele
GG: not at risk
individuals w/ risk allele have highest BMI, those w/o risk allele have lowest BMI
however, those who're carriers of 2 risk alleles benefit most from doing physical activity; those who're homozygous or heterozygous for WT, no risk allele, do not benefit from exercise, to lower their BMI
why is knowledge of genetic risk important/helpful?
1) can inform personalized medicine, leading to prediction of disease risk & prognosis or adapting therapy to individual patients
2) expands our understanding of biological pathways of disease, leading to development of targeted therapies