Automatic clustering for gene expressing matrix

Question

[J.F.Jiang]

Hi all,

I am dealing with the co-expression modules using WGCNA.

WGCNA is highly recommend for its softhresholding and preservation analysis of the expression data.

The method can automaticly generate some modules (networks) in co-expression manner.

Basically, the co-expression modules can be obtained as followed,

1. calculate the pearson correlation of the expression matrix
2. set a predefined cutoff (e.g., 0.8~0.9) to obtain those highly connected edges
3. apply k-means, SOM... clustering method to obtain the modules,

However, these clustering method required a given number for the expected clusters.

I am wondering if there is any tool that can take the expression matrix as input, and automatically obtain the co-expression modules?

Best

[roy.granit]

Maybe this tool can help you.

[J.F.Jiang]

Thanks, the method in the web you offered is a combination of several popular methods, such as kmeans,hclust, and so on.

What I mean is to find a tool that can handle expression matrix as input just as WGCNA.

Thank, if there is no one, I will write the script.

[PoGibas]

But why can't you use WGCNA?

[J.F.Jiang]

I just want to compare the outputs of them.

[PoGibas]

Does WGCNA (blockwiseModules) require number of modules?

[J.F.Jiang]

WGCNA will not require the input of module number, so it is "automatic".