Question

GeneSCF and P-value Benjamini and Hochberg (FDR)

0

Entering edit mode

6.7 years ago

aindap ▴ 120

Dumb question about the GeneSCF output. If I'm using the P-value Benjamini and Hochberg (FDR) column, would I choose an FDR cut point and take any rows with a BH P-value less than this value as significant?

Thanks so much.

GeneSCF • 2.2k views

ADD COMMENT • link updated 6.7 years ago by EagleEye 7.5k • written 6.7 years ago by aindap ▴ 120

score 2 · Answer 1 · 2017-08-03

2

Entering edit mode

6.7 years ago

halo22 ▴ 300

I haven't seen the GeneSCF output but since FDR is used for multiple correction and I assume you are comparing many genes it makes sense to use all rows with an FDR value < 0.05.

ADD COMMENT • link 6.7 years ago by halo22 ▴ 300

score 2 · Answer 2 · 2017-08-03

Hi,

In GeneSCF if you are choosing FDR cutoff of 0.05, then whatever entries less than 0.05 is significant and take those rows (Same applies if you consider only P-values for filtering).

My suggestion (Again it depends on your dataset/experiment): Since GeneSCF also considers terms containing entries with only one or two genes in the database, in addition to P-values/FDR I would also prefer to use minimum number of genes to consider ('num_of_genes' column) as filter criteria from the output.