Data Normalization For Co-Expression Network Construction By Wgcna

Question

[l0o0]

I get fpkm value from different traits of different tissues by tophat and cufflinks workflow. Then i rearrange the values, traits as row names and gene id as col names. I want to use this dataset to construct co-expression network by WGCNA.

I know the first step is to do data normalization. There are many normalization methods, which one do i apply to this dataset for WGCNA?

Answer 1

[Ashutosh Pandey]

FPKM data is already normalised. Just follow a tutorial (http://labs.genetics.ucla.edu/horvath/CoexpressionNetwork/Rpackages/WGCNA/#tutorials) that is most relevant to your problem.

[l0o0]

Thank you for your quick reply :)

Answer 2

[paul.e.gradie]

To bump an old question:

Data normalization is handled internally within Cufflinks on a PER SAMPLE basis. For each transcriptome you assemble or quantify, calculation of the FPKM values include a normalization step using the DEseq geometric mean normalization method.

If you want to compare across multiple samples which were quantified independently, then you need to normalize these to one another by using a tool such as CuffNorm. However, if you run CuffDiff - or any other tool that automatically normalizes any input data sets - then normalization occurs using the default geometric normalization method between the samples included in the analysis.