Hello to all,
I hope everyone is well! I want to ask if anyone knows if there are any parameters in STAR that prevent the alignment of reads to unannotated splice junctions? I want to speed the alignment task by limiting alignments to splice junctions that are not known. I already included --sjdbScore to 0, --alignSJoverhangMin to 10. But, it still takes some time and I still get splices that for the moment don't interest me.
Thanks in advance for your comments!
•
written
by
MarVi •
The parameter: --alignSJoverhangMin sets the:
minimum overhang for unannotated junctions
Whereas, the parameter: --alignSJDBoverhangMin sets the:
minimum overhang for annotated junctions
Note the presence of SJDB vs. SJ in the latter, this refers to your GTF database and/or 2-pass splice junctions, etc... Essentially, what you consider to be annotated.
In other words, for your specific case you should change --alignSJoverhangMin <N> so that N is a number greater than your read length.
You should no longer get any spliced reads to unannotated sites with overhangs shorter than N (which means no reads should splice to unannotated junctions).
Assuming standard RNA-seq you may be fine just setting it to something like 1000.
You can read more here: https://groups.google.com/u/1/g/rna-star/c/J6qH9JCysZw/m/VmgoGKSE9qQJ
If you want to modify the actual outputs in the splice junction files.
Log in to answer this question.
