Quantifying genes' expression with Trinity

Question

[langziv]

Hello again.
The RNA-seq data I got was obtained from an experiment with the following structure:
3 similar experiments in separate dates, each included a control group and an experiment group.
I thought that It would be reasonable to create one assembly for the 3 experiments' reads, and another for the 3 controls' reads, since an assembly from an experiment might include different RNAs than that from a control assembly. I'm not sure because I didn't see anything about that in the Trinity documentation.
If indeed 2 separate assemblies are is needed, quantifying genes' expression with the script align_and_estimate_abundance.pl from the "Trinity tool-kit" should be done twice: once for the controls and once for the experiments?

[biofalconch]

The flaw with your pipeline is that the two transcriptomes are not equivalent, and therefore cannot be compared as they are now. I think the easiest two solutions would be:

1) Reconstruct the de novo transcriptome using all the samples. It doesn't matter if some RNAs will be present in one and not the other, you just care about reconstructing the best transcript models as you can. The differences you can deal with at the quantification level. If your control samples has some RNAs not present in your experiment, it will be reflected in the expression levels.

2) You can try and unify the transcriptomes into one with CD-HIT or something of the sort, but this approach is too convoluted and probably not the best to deal with.