acesnik
commented
6 years ago There is a tradeoff here:
- Combining read files before the whole analysis is simpler to code and manage
- But processing the read files separately can give useful information regarding the consistency of variant detection. This is especially important for splice variants. We could use rMATs, which I've already started wrapping or JUM, which outperforms rMATS in number of interesting ways, to find which splice junctions are consistent between tech reps. This is an idea that comes from here https://www.biorxiv.org/content/biorxiv/early/2018/07/27/372995.full.pdf.
Alternatively, allow combining fastqs after downloading SRAs.