1
$\begingroup$

I am performing a GWAS analysis and try to understand the influence of the minor allele count filter.Setting the filter to 1 % gave me this plot and I am confused about the same -log10 pvalues around ~ 3.8 which shows a strange pattern (many dots in one line with same value).

enter image description here

If I increase the filter to 6 % the dots in a line disappear but I loose also some significant SNPs:

enter image description here

Sample are about 77000 SNP positions (barley), unimputed from GWAS (all NaNs removed). GWAS was performed with 200 phenotype samples.

How to interpret the dots in a line?

Improve this question
$\endgroup$
4
  • $\begingroup$ I updated my question. $\endgroup$
    – snowflake
    Commented Aug 21, 2020 at 6:33
  • $\begingroup$ Thanks. How many samples and what organism? I assume this is a snp chip? $\endgroup$
    – winni2k
    Commented Aug 21, 2020 at 6:44
  • $\begingroup$ Barley with 200 phenotypes. $\endgroup$
    – snowflake
    Commented Aug 21, 2020 at 7:04
  • $\begingroup$ There seems to be something wrong with the Manhattan plots: snps from different chromosomes are overlapping? For example, green does can be found inside the region for red dots. $\endgroup$
    – winni2k
    Commented Aug 21, 2020 at 7:23

1 Answer 1

Reset to default
1
$\begingroup$

This looks like genotyping failed for some samples. It's unclear from the op whether the genotyping data was controlled for quality, but I would recommend performing sample QC.

On a side note: in human gwas with imputed snps, I would expect to see clusters of significant snps as a sign that the significant snps are not an artifact. In the op there are some single significant snps. This might be a sign that those significant snps are artifacts. There are also some snps that form clusters and therefore look more promising.

Improve this answer
$\endgroup$
2

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.