Question

determine k during RUVSeq

0

Entering edit mode

capricygcapricyg ▴ 10

@capricygcapricyg-17892

Last seen 2.2 years ago

United States

Dear Support Team,

I have RNA-seq data from 50 samples. I would like to run RUVSeq to correct the unwanted variables. I wonder if there is a general guide about how to choose k number during RUVg? The online documentation used 1, but I saw some discussion about using other numbers...

Thanks a lot!

C.

ruvseq • 2.2k views

ADD COMMENT • link updated 5.8 years ago by davide risso ▴ 980 • written 5.8 years ago by capricygcapricyg ▴ 10

score 0 · Answer 1 · 2018-11-05

0

Entering edit mode

davide risso ▴ 980

@davide-risso-5075

Last seen 5 months ago

University of Padova

Hi,

we haven't found a good way to estimate K in RUV, but exploratory data analysis can be helpful to choose a reasonable value for K.

My answer to this thread may be relevant (minus the time course aspect): A: RUVseq: How to implement to more complex time-course model?

ADD COMMENT • link 5.8 years ago davide risso ▴ 980

0

Entering edit mode

Thank you for the quick response!

I saw an general increase in within sample dispersion on PCA plot with increased k. Also, when k=1, the within sample dispersion increases when compared with non-corrected sample. Is it normal?

Thanks a lot!

C.

ADD REPLY • link 5.8 years ago capricygcapricyg ▴ 10