STAAR sliding window errors about staarpipeline-tutorial HOT 4 CLOSED

Dear Alex,

Thank you for letting me know. We have yet to encounter this type of error. After a brief investigation, I can confirm that (1) all of the results that you currently have (the p-values in your Manhattan plot) are valid; (2) this is more of a numerical issue, which relatively small sample sizes may cause.

We will take a further investigation and see if anything can be amended to avoid this error.

Best,
Xihao

from staarpipeline-tutorial.

Dear Xihao,

Thank you for replying. I'm working with ~400 cases and ~700 controls, so indeed my sample size is considerably smaller than the example in your paper (~20,000).

The default setting for the sliding window analysis is a window size of 2 kb and a "rare_MAF_cutoff" of 0.01. To try to circumvent my small sample size issue, I increased this to a window size of 100 kb and a "rare_MAF_cutoff" of 0.05. However, it still doesn't work.

With 800 alleles in cases, I should be able to detect variants down to a frequency of 0.125%. The rare MAF cutoff is above this, at 5%. In addition, I am accumulating alleles across a window of 100 kb, which is a reasonably large distance.

So, somehow, I am not able to detect at least two variants less than 5% MAF in the majority of 100 kb windows across the genome, using 1100 individuals.

Also, I'm not sure a low sample size explains why the sliding window works properly only at the start of chromosomes?

Do you have any suggestions? For instance, what parameters I should choose for the sliding window analysis? Perhaps I should try the dynamic window analysis?

Many thanks for your help,

Alex

from staarpipeline-tutorial.

Dear Alex,

Thank you for providing the additional information. Your justifications for using a "rare_MAF_cutoff" of 0.05 and longer sliding windows are reasonable. Yes, please try the dynamic window analysis, and feel free to keep us posted via email if you have any follow-up questions.

Hope this helps.

Best,
Xihao

from staarpipeline-tutorial.

Thank you, Xihao, for your feedback.

from staarpipeline-tutorial.