RNA-seq calling pipeline feedback

[gaow]

Here are some useful feedback from discussion with GCAD today:

Major

1. We should improve our report at the end of each QC, not just how many sample/gene etc we remove but also what they are. Because perhaps some researchers don't like the way we do QC for their specific applications. In that case they can review our report, but start from the preQC data themselves. Of course we still provide QC-ed files to give people opportunity to fiddle with them and decide -- we are expected to receive a list of QC metric to report from GCAD (we should have the information but need to formalize the report)
2. We should add a feature to automatically detect library type for use with adapter trimming, not relying on user's knowledge or guess. They suggested TrimGalore to detect it. I have not looked into this yet.
3. To detect RNA-seq protocol stranded vs unstranded we use infer_experiment.py from RNA-SeQC and this should be done before we prepare the reference data because the reference preparation has to do with this protocol type.

Minor

1. Software version: should we update to the latest in case some bug fix matters? eg RNA-SeQC is on 2.4.2 in our image but there is 4.0.0 release. We should study their release note. I don't want to update without a good reason