STAR 2-pass mapping with multiple conditions

[Samuel Zimmerman]

Hi all,

In order to do multi-sample 2-pass mapping the recommended approach is to collect 1st pass junctions for all samples. Then use all the junction files for mapping each sample in the second pass.

My question is, if I have 2 conditions,  (4 biological replicates of each for a total of 8 samples) should I only use the 4 junction files created with the samples for my wild-type condition during the 2nd pass of wild-type samples, or should I use all 8 junction files on every sample regardless of the condition?

Also, on a separate note, does it make a difference in terms of quality and speed when you include the annotation GTF file, during the genome generate step or during the mapping step?

Thank you very much

Best,
Sam

[Alexander Dobin]

Hi Sam,

I would recommend including junctions from all 8 samples. This achieves the highest sensitivity to novel junctions.

For instance, if there is a novel junctions that's is highly expressed (many reads, confident detection) in the wild-type, but only weakly expressed (few reads) in the experimental group, using junctions detected in all samples for the 2nd pass of all samples will allow to detected those few spliced reads in the experimental group.

Cheers

Alex