davidaknowles / leafcutter

Annotation-free quantification of RNA splicing. Yang I. Li, David A. Knowles, Jack Humphrey, Alvaro N. Barbeira, Scott P. Dickinson, Hae Kyung Im, Jonathan K. Pritchard
http://davidaknowles.github.io/leafcutter/
Apache License 2.0
204 stars 115 forks source link

Use of IntronMotif in STAR #169

Open OgnjenMilicevic opened 3 years ago

OgnjenMilicevic commented 3 years ago

In your paper and documentation you kindly show your command lines. When using STAR, you use the option --outSAMstrandField with IntronMotif which according to STAR manual: "The spliced alignments that have undefined strand (i.e. containing only non-canonical unannotated junctions) will be suppressed."

Can you elaborate on the importance of this option? STAR manual says XS tag is only for Cufflinks compatibility. Is it essential for leafcutter? If we are searching for annotation-free results (as advertised), is it weakened by the filtration of unannotated junctions?

jackhump commented 3 years ago

Sorry for the delay.

Hi, I believe the intronMotif option will keep novel junctions, as long as they have a defined strand. So yes this would remove non-canonical unannotated junctions, but most novel junctions have canonical splice sites so will be kept.

I also believe that the XS tag is used by regtools in assigning strand to the junctions but I'm not completely sure. There's been a few recent bugs with regtools setting "?" for strand and I worry this is due to a dependency on the XS tag.