search

caitiecollins / treeWAS

treeWAS: A Phylogenetic Tree-Based Tool for Genome-Wide Association Studies in Microbes
Other
92 stars 18 forks source link

How can I get accurate P.value #29

Closed zhaoxvwahaha closed 5 years ago

zhaoxvwahaha commented 5 years ago

Hi,

The treeWAS is amazing tool, I use it in my data set now, but I have some questions about the p.value:

(1) I got too many significant markers from treeWAS, I wonder which are the most significant ones, i.e. I need the priority of these markers, but all the p.value=0 in output, so how can I get the accurate p.value rather than 0. (2) What's the min.p.value means, is it the threshold of bonf corrected p-value? (3) What's the p.value in out object of treeWAS (eg. $out$terminal$p.vals), they are base p-value or bonf corrected p-value. (4) I found that not all markers with 0 p-value (eg. $out$terminal$p.vals) are significant, i.e. the p-value of some markers are 0, but they are not significant.

Looking forward to your help, Thank you in advance.

zhaoxvwahaha commented 5 years ago

I'll appreciate it very much if you could help me about my above questions. @caitiecollins @xavierdidelot

xavierdidelot commented 5 years ago

1) You could try increasing the value of n.snps.sim to build a null distribution using more simulated sites and therefore get more precise estimates of the p-values 2) min.p.value is the smallest possible value of p-value. So any value returned as zero should be interpreted as being in fact below min.p.value 3) they are corrected p-values 4) this may be because not all significant sites are printed on screen, but they are still there in the output