Dear David,
Thank you very much for you help, I really appreciate it.
I am not using the read.snps.long() or any other import function, as the
data is already in snpMatrix, so I cannot specify it at the input stepâ¦
I am reading the data as a snpMatrix, so using load() after having called
the snpStats package as in the vignette: "Example of genome-wide
association testing":
require(snpStats)
data(for.exercise)
The objects loaded with this set are: 1) genotypes (in probability format
from imputed data); 2) SNP.support object, with information on the SNPs
such as Allele 1, Allele 2, chr, position. Other information that I need
for the meta-analysis can be extracted from the 'col.summary' command in
snpStats: MAF, and I think RAF (risk allele frequency) can be considered as
the Effect allele frequency for the meta-analysis.
Then I am using snp.rhs.estimates and snp.rhs.tests for the associations.
The problem is that I don't know which allele is taken as the risk allele
in the association, is there a way to see this? Is it always the Allele 2
reported in the SNP.support file? Until I understand this, I won't be able
to harmonise the SNPs all to one reference, since I don't know if I need to
flip the betas when the effect allele is reversed for exampleâ¦
I have noticed this because when comparing the frequency of the Allele 2
(taken as the risk allele) and the RAF which I thought was the frequency
associated with it, with the frequencies of the same allele found in the
1000 Genomes, I get concordance up to frequency= 0.5, then a shift in
direction happens and I get discordance up to 1 for the reference
frequency.
Thank you very much for any suggestions you may have,
Francesca
2014-05-27 2:07 GMT+01:00 David Duffy <david.du...@qimr.edu.au>:
> On Mon, 26 May 2014, francesca casalino wrote:
>
> I am having this problem because I need to run a meta-analysis and to
>> align
>> all the variants between the different studies included in the
>> meta-analysis I need to know the effect allele used to get the beta (so
>> that I can flip the beta if the effect allele is flipped compared to all
>> other studies).
>>
>
> This depends on how the data has been sent to you. Obviously, you should
> check the "A" allele frequency in the different datasets. If they have
> used different genotyping assays and the strand of the SNP is ambiguous, eg
> G->C transversion, then this may be the only way to exclude problems. PLINK
> offers a tool to check for this using LD patterns.
>
> Cheers, David.
>
>
> | David Duffy (MBBS PhD)
> | email: david.du...@qimrberghofer.edu.au ph: INT+61+7+3362-0217 fax:
> -0101
> | Genetic Epidemiology, QIMR Berghofer Institute of Medical Research
> | 300 Herston Rd, Brisbane, Queensland 4006, Australia GPG 4D0B994A
>
[[alternative HTML version deleted]]
______________________________________________
R-help@r-project.org mailing list
https://stat.ethz.ch/mailman/listinfo/r-help
PLEASE do read the posting guide http://www.R-project.org/posting-guide.html
and provide commented, minimal, self-contained, reproducible code.
