Too many significant results

[Lahat A]

Hi,
My tabulated results show about 8400 genes/events. However 8000 of them have p values bellow 0.05 (I averaged “probability_changing”). Having so many results makes it impossible to do further analysis such as reactome enrichment, and it seems suspicious to me to have so many splicing events.

 

Could it be that I am doing something wrong? is there a better way of filtering them?

[Paul Jewell]

Hello,

Can you clarify the run arguments? Are you using the .tsv file output by "majiq psi/deltapsi/het" or the one from "voila tsv" ? The latter option provides a large number of comprehensive filtering options to reduce the output size. Check "voila tsv --help" to see them all. For example, you may use the switch "--changing-pvalue-threshold" to set a higher threshold for those rows.

[Joseph Aicher]

probability_changing implies that you are using majiq deltapsi. These are not p-values, but rather posterior probabilities. That most events/junctions have low probability of changing (abs(dpsi) > threshold) would suggest few significant results, not too many.

Best,

Joseph

--
You received this message because you are subscribed to the Google Groups "majiq_voila" group.
To unsubscribe from this group and stop receiving emails from it, send an email to majiq_voila...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/majiq_voila/79016fd8-f379-40d0-bba2-6a06960974f2n%40googlegroups.com.

[Lahat A]

I am using the tsv that comes out of delta psi. I will check the 
output of "voila tsv" next. thanks.