I have rna-seq data from two different tissues and different stress conditions. I have performed adapter removal using Cutadapt and I have used TopHat for alignment to the reference genome. What is the ideal method to perform differential gene expression analysis? Should I perform CuffDiff on individual stress-tissue dataset, extract the FPKM files from each of the experiments and merge them using some normalization method ? I need a way to normalize the read-counts across all the stress-tissue datasets because I want to construct co-expression network using RNA-seq data. Any suggestion will be very helpful. Thanks.