Error when calling module_stability for networks with cor_method biweight

[jucapitanio]

Hey all, great package! Thanks for putting this together.
This is a very simple bug. When calling module_stability on a network created with cor_method = biweight,
line 7: cor_method <- net$params$cor_method
will be biweight and when should be bicor when you try to pass that to WGCNA here:
lines 10-15:

mods0 <- WGCNA::sampledBlockwiseModules(nRuns = nRuns, replace = FALSE, 
    datExpr = expr, maxBlockSize = 5000, checkSoftPower = FALSE, 
    corType = cor_method, networkType = net_type, TOMType = TOMType, 
    TOMDenom = "mean", mergeCutHeight = module_merging_threshold, 
    reassignThreshold = 0, numericLabels = FALSE, checkMissingData = FALSE, 
    quickCor = 0, verbose = 2)

Of course we can just change that with net$params$cor_method <- "bicor" before running module_stability, but I figured I'd let you know anyway.
Thanks again for creating this great resource.

[almeidasilvaf]

Hi, @jucapitanio

Thanks a lot for the positive feedback and for reporting this issue.

Based on your workaround, the issue can fixed by adding a single line of code:

cor_method <- ifelse(cor_method == "biweight", "bicor", cor_method)

I'll let you know once I push an updated version with the bug fix.

Best,
Fabricio

[almeidasilvaf]

Fixed in version 1.9.2.