Page 1 of 1

Association tests in families and unrelated

Posted: Wed May 16, 2012 9:28 am
by fede
Hi,
I don't know if this question is appropriate for this forum.
I would need a suggestion.
I should run a linear regression on a quantitative trait for around 15000 SNPs in a sample including families and unrelated individuals.
I wanted to try the q-fam total (family based association tests for quantitative traits --qfam-total) Plink package ( I never used this before). I started my analysis but I discovered with this package I can't use covariates.
Do you ever used it?
Do you have any other suggestions to perform my analysis?
Thanks a lot

Federica

Re: Association tests in families and unrelated

Posted: Mon May 28, 2012 8:17 am
by Nicola Pirastu
Hi,

I never used plink to run such analises, however I have a similar population (mixed families and unrelated) and use GenABEL with mmscore.
Everything works fine for me so I would advise you to use the same approach.

You can find all the details in the GenABEL tutorial.

Best

Nicola

Re: Association tests in families and unrelated

Posted: Thu Jul 12, 2012 6:46 am
by fede
Thanks a lot!!!!
I lost your answer :(
I'll try now and let you know :)
Cheers

federica

Re: Association tests in families and unrelated

Posted: Thu Jul 19, 2012 7:33 am
by fede
Hi Nicola,
thanks a lot for your advice.
I read the tutorial and run this:

1. Create tped and tfam from plink binary files
plink --bfile filename--recode --output-missing-genotype 0 --transpose --out outputfile

2. Import tped and tfam in GenABEL format
convert.snp.tped("outputfile.tped","outputfile.tfam","mydata.raw",strand = "u", bcast = 10000)

3. Load files
dataset <- load.gwaa.data(phe = "pheno.txt", gen = "mydata.raw")

4. Analysis
gkin <- ibs(dataset,w="freq")
df <- dataset[,autosomal(dataset)]
names(phdata(dataset))
h2ht <- polygenic_hglm(eGFR_all ~ sex + sbp + dbp + age + bmi + therapy + centre, kin = gkin, df)
mms <- mmscore(h2ht,data=dataset)

5. Descriptive
descriptives.scan(mms, sort= "Pc1df", top=11000, file="Results.csv")

6. To get P- HWE, Q.2 and CallRate
additional_data <- summary(dataset)


It looks like it worked :) :)
cheers
fede

Re: Association tests in families and unrelated

Posted: Tue Aug 21, 2012 9:01 pm
by Nicola Pirastu
If you are interested in all those summary statistics you should try also the formetascore() function,
it will format everything very neatly.