Question: Plink Family Structure/Population stratification
1
gravatar for Tom
5.9 years ago by
Tom40
United Kingdom
Tom40 wrote:

I'm wondering if someone could please help. 

I have a population of individuals, some are in families and some are not in families. I have a binary phenotype. I want to do an association analysis between phenotype and genotype, controlling for population stratification and family structures in the population (and other co-variates that I have in the --covar file).

The command I'm using is: 

plink --bfile myfile --logistic --ci 0.95 --out TestPlink --noweb --covar CovariateFile --covar-name SEX,AGE --hide-covar --model-rec 

First, I'm checking for population stratification using the command: plink --bfile MyFile  --ibs-test  --out TestPlink --noweb. This is currently running.

Second, I want to account for the family structure of my population. I know for sure that my population is made up of families (i.e. it was a community based sample of maybe 300 families + individuals not in families). I've tried to read thru the doc, and also other papers that use PLINK, but I'm just really confused on what the command is that says "look in my population for evidence of family clustering, and account for this family clustering when doing the association analysis". For example, do I use permutation "controlling for identified substructure or familial relationships by permuting only within cluster" , multidimensional scaling plots or the DFAM procedure in PLINK (implements the sib-TDT and also allows for unrelated individuals to be included (via a clustered-analysis using the Cochran-Mantel-Haesnzel)).

 

Would the correct command for "look in my population for evidence of family clustering, and account for the fact that there are families in my population when doing the association analysis" be something like this:

plink --bfile myfile --logistic --ci 0.95 --out TestPlink --noweb --covar CovariateFile --covar-name SEX,AGE --hide-covar --model-rec  --dfam 

And then would the command to also control for population stratification, should there be any be:

plink --bfile myfile --logistic --ci 0.95 --out TestPlink --noweb --covar CovariateFile --covar-name SEX,AGE --hide-covar --model-rec  --dfam --cluster (--mds-plot 4 ???)

 

Many thanks

ADD COMMENTlink modified 5.9 years ago by chrchang5237.6k • written 5.9 years ago by Tom40
0
gravatar for chrchang523
5.9 years ago by
chrchang5237.6k
United States
chrchang5237.6k wrote:

The standard procedure today is to use a linear mixed model to correct for population stratification and relatedness; a good survey paper is http://www.nature.com/ng/journal/v46/n2/abs/ng.2876.html .  PLINK does not currently implement this, but don't worry, most programs which do implement linear mixed model association are designed to operate on PLINK-formatted data.

Some noteworthy programs in this space:

GCTA: http://www.complextraitgenomics.com/software/gcta/

FaST-LMM: http://research.microsoft.com/en-us/um/redmond/projects/mscompbio/fastlmm/

BOLT-LMM: http://www.hsph.harvard.edu/alkes-price/software/

ADD COMMENTlink modified 5.9 years ago • written 5.9 years ago by chrchang5237.6k
Please log in to add an answer.

Help
Access

Use of this site constitutes acceptance of our User Agreement and Privacy Policy.
Powered by Biostar version 2.3.0
Traffic: 2416 users visited in the last hour
_