In my module to trait correlations analysis with WGCNA I find significant correlations in several cases. I do however also see correlations with the grey module is the range of about 0.5 - 0.6. The grey module consists of 8390 genes out of the total 14729 in my dataset. This leads me to belive that there is an overall correlation between the dataset and the traits. If this is the case then would in be more interesting to see which modules have the opposite correlation to that of the grey module and a specific trait?
Is this a correct assesment and how should I be thinking about this -- is there anything Im missing?
1 answer
If this is the case then would in be more interesting to see which modules have the opposite correlation to that of the grey module and a specific trait?
I do not know why this would be of interest, after taking into account the reason why the grey module exists in the first place. Also, these correlation coefficients do not say much about whether or not the correlation is statistically significant.
With the data that you have, you should focus on the statistically significant correlations to the non-grey modules, and then, for example, perform gene enrichment on the genes in each. By going back a few steps and modifying parameters, you can likely increase or decrease the number of genes assigned to grey.
Kevin
Log in to answer this question.
That is entirely possible because the clustering algorithm is not perfect, so it is possible that the hub genes of the grey module shows some degree of correlation with the hub of a different module. By looking at the heatmap of each module usually helps to understand how your gene/samples behaves