Hi, For input data for WGCNA, can I input all my data including the replications for each samples, or use average of the replications for each sample? Also to have a better chance to find modules responsible for a difference between female and male datasets, is it better to filter the data by COV to reduce number of genes in the input file? Thank you,
Hey there, you can input all data together; so, include the replicates.
Yes, you can pre-filter by CoV prior to running WGCNA, if you wish. This may make the computational time shorter, but it should be pointed out that WGCNA produces a 'weighted' network, and would therefore likely assign lesser weights to edges / connections of little variation anyway.
Kevin
Use of this site constitutes acceptance of our User Agreement and Privacy Policy.
version 2.3.6
Got it, thank you very much
Sorry can I ask you another question? After running WGCNA, how can I know that my modules and network is good enough? In my case I get 12 modules, some of them with positive and some other with negative effect and also some other with almost no effect on the phenotype. Is this what we expect? what else, other than power, I can change to get better modules? Thank you,
I am not sure how to define 'good enough', but, prior to obtaining the final results, a few different QC graphs should have been generated - you could check these. Also, prior to even running WGCNA, you should obviously filter your input data for, e.g., lowly expressed variables and those that have low or no variance.
Awesome, thanks a lot