highly correlated hidden factors

0

Entering edit mode

2.8 years ago

GK1610 ▴ 110

I have a matrix of ATAC-Seq counts using imputation. I called hidden factors then using PEER bayesian method. The problem is all PEER factors are highly correlated like spearman correlation > 0.9. Does anyone has the feeling why it is happening?

ATAC-Seq data exploration analysis • 781 views

ADD COMMENT • link updated 2.0 years ago by Angli Xue • 0 • written 2.8 years ago by GK1610 ▴ 110

0

Entering edit mode

Hi, have you solved this problem? I met the similar situation and trying to find out why.

ADD REPLY • link 2.2 years ago by Angli Xue • 0

0

Entering edit mode

use scale function in R to properly normalize the input matrix before calling PEERs and the problem of correlation will go away

ADD REPLY • link 2.1 years ago by GK1610 ▴ 110

0

Entering edit mode

Could I ask what is your sample size? I found that the correlation is still there when the sample size is ~30.

ADD REPLY • link 2.0 years ago by Angli Xue • 0

Login before adding your answer.