Posotive controls for GWAS results
Hello,
Is there is a way to check GWAS results for a particular phenotype aside looking up individual SNPs for known associations in https://www.ebi.ac.uk/gwas/search?query=rs6742078
Thanks
• 135 views
•
link
0 answers
No answers yet.
Log in to answer this question.
Hey again, you may have to elaborate a bit on what you are aiming to do. You have a set of SNPs that pass genome-wide statistical significance and you want to see to which phenotypes they have been previously associated?
You can download the GWAS Catalog and use that in some way - it would be quite informative: https://www.ebi.ac.uk/gwas/downloads
Hey again:) I am repeating GWAS which was done 10 years ago and results of that are published on that GWAS Catalog. The issue is that I am not getting the same results...even when I am doing GWAS with the most basic plink --assoc which was probably what was done 10 years ago. I am wondering if my new GWAS "makes sense" and if there is a way for me to justify that what I am doing now is correct. I am doing it for diabetic retinopathy EUR for type 1 diabetes subjects and that GWAS Catalog features studies that that don't comply with all those requirement aside the above mentioned old study. I found this software: https://cran.r-project.org/web/packages/QCGWAS/QCGWAS.pdf
but there I don't know how to make plink2 output me these parameters that are needed to run QCGWAS: STRAND, HWE_PVAL, CALLRATE, IMPUTED, USED_FOR_IMP
Right now I am running it like this:
Any ideas?
The fact that it fails to even replicate on the same dataset is not surprising, but in which way does it not replicate? Are the Odds Ratios inverse or is it just that the p-values are different? QC filtering is likely also an area at which to look.
From what I know of diabetes and DR, there are going to be [possibly] many thousands of genetic variants that increase risk.
Also, do you need to use QCGWAS or can you just generate some summary QC on your own?
Kevin
The P values are different, I am not getting the signal in the same loci, neither on the loci that are in LD with the previously reported significant SNPs
I suppose that is at least some form of replication! I am not sure why this would happen, but different software versions and both bug fixes and altered default values may explain it.
And different imputation platform...Thanks...I just wonder how people replicate previous studies these days...
GWAS is infamous for its poor replication... well most of research is, to be honest
How about: 1. taking a set of common SNPs for one person and calculate correlation between allele frequencies for old GWAS and for new GWAS and compare that. 2. And than doing the same for all individuals and create two distributions and than compare those?
Do you know any software or workflow which I can use to do 1?
right now I only know how to do that for all subjects and extract the SNPs which are common via:
plink --bfile Merge --freq --out freq_stat a=fread("freq_stat.frq") b=a[a$MAF<=0.05,]
but how to calculate these allele frequencies per subject?
The allele frequency per subject can only be 0, 0.5, or 1.0 for a bi-allelic site (?)