Hi everyone,
I am carryied out a GWAS with the sample size about 5000 control versus 5000 cases, and 260,000 variants at the beginning
After the quality control with MAF at 0.05, I achieved only 36,000 variants (without MAF, I still have about 150,000 variants). I have checked the frequency of variants and found that most of them have really low MAF, that is the reason why I have so low variants after QC (just 36,000).
I have draw a QQ plot for the data ( 주석 2019-12-07 140011.png ) and see that the QQplot is not as expected. With the result as this QQplot, a material had recommended me to find an analysis strategy. What should I do in this case? I am planning to do imputation and check again, but can imputation can help to solve the problem?
To start please don't use so many things GWAS and MAF. I assume this means Genome wide association study and Minor Allele Frequency.. It is quite helpful to say what you are plotting on your axes. Not all Q-Q plots are the same. Given that you don't say what you did in the of experimentation or data analysis I will direct you to a paper that I hope will help:
Article Voorman A, Lumley T, McKnight B, Rice K. Behavior of QQ-plot...
This and other useful things can be found here:
https://www.google.com/search?rlz=1C1CHBF_enUS874US874&ei=u03yXb-UHIaQtAb856LoAQ&q=GWAS+MAF+Q-Q&oq=GWAS+MAF+Q-Q&gs_l=psy-ab.3...576197.579875..581836...0.2..0.96.359.4......0....1..gws-wiz.......0i71j0j0i22i30j33i160.oRlOMYqlkL4&ved=0ahUKEwj_3eaUqLDmAhUGCM0KHfyzCB0Q4dUDCAs&uact=5
another helpful link:
https://stats.stackexchange.com/questions/185284/deflated-qq-plots-in-genome-wide-association-studies
Please next time say what you did, how you did it and what you found. Then just keep SNPing away at it. Sorry but I like puns. Best, D. Booth
Hi
Your problems come from data structure, you must check your MAF for each SNPs. To do this you could carry out different MAF and check the results. How many SNP remain after different QC. also check the position of your SNP (SNP distances ) may couse problemsin case such as LD analysis, maybe GWAS.
Thank you so much Professor. I am so sorry when I did not mention clearly.You are right that I am checking the Genome wide association studies. Back to the problem, I have checked the Minor Allele Frequency (MAF) and I would like to show you that this is the MAF plot. I think the problem is because most of the variant has too low MAF (most of them are 0) hence in the MAF step, all of them was excluded (only 36000 variants remained). The QQ plot was drawn based on the "qqman" package in R and I used the p-value of logistics method (-losgistic in plink). I also check when I used the p value of some other adjustments (False discovery rate; Benjamini & Hochberg (1995) step-up FDR control; Sidak single-step adjusted p-values... The qq plot looked the same as my previous one (the change can be neglected).
- Badges
- Science method