Question: WGCNA with time series data
gravatar for julia
6.8 years ago by
julia20 wrote:

I have been going through a set of tutorials on WGCNA for building weighted gene co-expression networks. One particular tutorial uses Yeast expression data collected over 44 time points across cell cycles. To create an adjacency matrix, a Pearson correlation is calculated for each pair of genes using expression over time. However, as I understand it, a Pearson correlation is not appropriate for time series data because the data are correlated over time periods. I just wonder if anyone has an explanation for why it would be appropriate to use a Pearson correlation with Yeast time series data. I would like to use WGCNA to analyze time series data in which expression values are obtained at multiple time points in the same patient, and I need to be able to justify this. Thanks for your thoughts on the matter.

rna-seq next-gen R gene • 5.2k views
ADD COMMENTlink modified 3.9 years ago by Biostar ♦♦ 20 • written 6.8 years ago by julia20

Hi Julia, I have a similar question on my time course data. My thought is using DayX-Day0 data instead of DayX would be beneficial for avoiding the variations from individual baseline background (Day0). Do you think it's making sense? How did you do in your study? Thanks.

ADD REPLYlink written 23 months ago by joseph.houjue0
gravatar for Ying W
6.8 years ago by
Ying W4.0k
South San Francisco, CA
Ying W4.0k wrote:

I am assuming that you are referring to the Steve Horvath's Yeast tutorial. In this tutorial, Pearson's correlation is run on the different eigengenes (datME) and not directly on gene expression over time. You obtain correlation between the various module eigengenes and cluster based on that.

ADD COMMENTlink written 6.8 years ago by Ying W4.0k

Thank you for responding to my question, Ying. Yes, I was referring to Steve Horvath's Yeast tutorial. Very early in the tutorial, prior to the identification of eigengenes, an adjacency matrix is created using the WGCNA adjacency function ( This function uses a Pearson correlation to create the matrix and transforms r to a connection strength measure using a power function. In the case of the Yeast data, the expression values across time are correlated for every pair of genes in the dataset. I am interested to know if there is a reason why a Pearson correlation would be justified in this case. I would really like to use this analysis with time series data.

ADD REPLYlink written 6.8 years ago by julia20

Could you please provide the link to this tutorial? I cannot find it.

ADD REPLYlink written 3.0 years ago by tlorin310
Please log in to add an answer.


Use of this site constitutes acceptance of our User Agreement and Privacy Policy.
Powered by Biostar version 2.3.0
Traffic: 1869 users visited in the last hour