Concatenate RNA-seq technical & biological reps or just technical reps before aligning

Question

[arfesta]

Say my goal is to identify SNPs or Differentially expressed transcripts between 3 main groups:

 1. High Fat
 2. Medium Fat
 3. Low Fat

For each group I have 3 biological reps (3 distinct progeny from a specific cross) and 2 technical reps of each biological rep.

Example:

High Fat contains 3 biological replicates of a cross between Parent_1 x Parent_2

Medium Fat contains 3 biological replicates of a cross between Parent_3 x Parent_4

Low fat contains 3 biological replicates of a cross between Parent_5 x Parent_6

Each of these biological replicates has 2 technical replicates.

I know that standard protocol suggests concatenating the technical reps and then aligning (and eventually merge biological reps with read groups), but if my objective is to detect SNPs or DE transcripts between the 3 main groups (i.e. the 3 distinct families which represent each group) should I consider concatenating biological and technical reps before aligning?

Possibly a poor question, but would appreciate any feedback.

[TriS]

for DE transcripts keep the biological replicates separated, align them separately and then calculate DETs/DEGs. for SNPs I'd suggest to do a similar approach. call SNPs per each replicate and keep those common across biological replicates.

Answer 1

[h.mon]

Why did you do technical replicates? In general they are not needed. Anyway, as you made them, check if they can be merged first, that is, no batch effect of any kind - NGS technical replicates should be highly similar (until they aren't).

For differential gene expression, biological replicates should always be separate - that is the point of biological replicates, give a measure of the variance of your population.

For SNP calling, you should not merge your biological replicates, as they will most likely differ at some SNPs, even if the parental lines are isogenic (which you didn't mention), as even isogenic lines have some variation. In case your parental lines are not isogenic, then the progeny genotypes will surely differ - due to recombination and meiotic segregation - and there is no reason to discard disagreeing SNPs across biological replicates.