W7L2 Thurs human population genetic P4

Question 1

genetic backgroud

Answer 1

-human genome consists of 3 billion base
-chromosome taken from different people are very similar. similarity up to 999/1000
-a significant part of variation in physical characteristic and susceptibility to diseases is associated with these genetic differences between individuals

Question 2

How to look at Genomic data

Answer 2

-it is straight forward to measure SNPs, position in the genome where it is know in advance that chromosome in the population carry one of two possible nucleotides

Question 3

Genotyping array

Answer 3

-typical modern human data sets contain thousands of individuals each typed at alot of SNPs
-cost 50 to measure up to 1M SNPs in a single human individual on genotyping array
-only measure a small fraction of your genome, but at places where population is known in advance to be variable
-miss rare and complex genetic variation

Question 4

cause for Pattern of genetic variation

Answer 4

-mutation (drift or positive selection)
-Recombination
-SNP data has a complex correlation structure. SNPs nearby each other on the same chromosome are typically inherited together, so genotypes at these SNPs can be highly correlated (linkage disequilibrium)

Question 5

Geographic differences

Answer 5

-individual who live near each other tend to be more similar genetically than individual who live in different part of the world for several reasons. one of these are due to shared genetic ancestry
-in genetic, this is called geographic population structure
-there can be very subtle difference in pattern of genetic variation even within country
-these pattern is useful to learn about the history of the population within the country
-also important to study genetic basis of disease

Question 6

geographic population structure study in UK

Answer 6

-UK population structure was studied in large disease association studies to check that it would not compromise analysis
-with the exception of a few particular, small region of the genome and a slight south north gradient in allels frequency, any population structure in the UK was quite subtle

Question 7

Previous approaches in detecting structure

Answer 7

-previous method ( structure and PCA) can detect population structure at continental or country scale
- implicitly ignore the correlation between nearby SNPs
-typically thin the data to obtain a set of SNPs which are close to independent of each other
-treating SNPs marginally throw away alot of information

Question 8

disadvantage of structure

Answer 8

-hard to interpret, especially for admixed sample
-does not cope well with continuous differences in allele frequency

Question 9

disadvantage of PCA

Answer 9

-hard to interpret, especially without labels
-cant assign individuals to cluster
-influenced by sampling biases and the projection can gave an undetermined effect on the conclusion one may draw

Question 10

fineSTRUCTRE: the new approach

Answer 10

-models the correlation between nearby SNPs (add power to detect subtle structure)
-no need to thin data
-explicitly assign individual to cluster
-determines the optimal number of cluster in the data in a principled manner
-capture a lot of information about shared ancestry
-record how often any individual in a sample has any other individual as its nearest relative

Question 11

fineSTRUCTRE: exploiting correlation

Answer 11

-estimate the genetic variants on the two separate chromosomes within each individuals
- for each individual, determine the number of chuck copied from each other individual
-treat these chuck vector as summaries for each individual, and use a clusterin algorthim to partition individual into cluster

Question 12

subtle population structure

Answer 12

• a powerful means of detecting population structure is to control and document carefully the provenance of the sample involved

Question 13

The POBI sample

Answer 13

-4371 samples
-fit the criteria of 4 grandparent born wihtin 60 km of each other, from rural area
-2886 individual genotyped at 600K SNPs
-2039 past QC
-cluster sample with finrSTRUCTURE

Question 14

fineSTRUCTURE and POBI

Answer 14

-application of fine structure break the POBI sample into 53 cluster
-fix the cluster and draw a tree by successively joining pair of cluster
-plot sample on a map and see how genetic clustering reflect geography

Question 15

Observation from fineSTRUCTURE and POBI

Answer 15

-for most part, the inferred cluster supported by historical , archeological and linguistic evidence
-shed light on archeological debate

Question 16

genetic composition of different groups

Answer 16

-aim to use a set of sample from different European population to ask how various UK cluster differ in term of their genetic composition
-6209 sample from ten country
-sampled as part of a diseas study: only have sampling location hosital or city
-cluster with fineSTRUCTURE

Question 17

Shared ancestry UK and europe

Answer 17

-use similar method to fineSTRUCTURE, but this time considering shared pattern of ancestry between UK cluster and European groups
-determine the average sharing of ancestry between UK cluster and European group
-Use these pattern of shared ancestry to learn about migration and population histroy

Question 18

Saxon admixture or replacement theory

Answer 18

-when the anglo-saxon arrive in England, there was a change in the language, in place name, crop and pottery style
-Some archeologist have speculated that the Anglo-Saxon replaced the local Romano-British population, unresolved
-the large group in central and southern england has substantial components of DNA from the Anglo saxon and more from ancient population

Question 19

Conclusion from the UK study

Answer 19

-region of UK not carrying genetic material from the anglo-saxon migrant
-genetic of saxon migration to be modest in England (at most 50%)
-significant pre-ROMAN migration into South East England
-ancestry of Orkney is only 25% Norse
-there are more sub group in UK than just Celtic population, some have distinct ancestry from each other based on our analysis