"bioinformatically pooling" replicates for SNP discovery from Pooled RNA-seq

[jfo]

Dear all,

What I have:
3 populations duplicate, 25M 100-PE reads each. That will be 6 libraries (using sample > 40 per lib) all in all.

Objective: discover SNP and use it for SNP genotyping and pop gen studies

Question:

Should I just combine Pop1_replicate1 and Pop1_replicate2 libraries (e.g., concatenate the two read1's and two read2's to get finalread1 and 2) before doing alignment, sorting, and SNP discovery? Any thoughts?

Thanks!