I am doing coexpression analysis on some data. I have around 7000 genes and 500 samples. I am using the step-by-step network construction method.
The number of modules change somewhat a lot when I use the deepSplit=2 vs deepSplit=4, I get 3 modules when I use the first and 13 when I use the second.
Moreover, when I plot the data using igraph, there are clearly more than 3 clusters that show up. I am afraid of grouping of the genes in the wrong modules. Is this much difference to be expected?
Is there a way to check the clustering of the genes is done correctly and if I should continue using the new sensitivity in deepsplit=4.