alignment step before using majiq

[Rotem Zilka]

Hello, 

I'm planning to use MAJIQ for alternative splicing analysis, and I want to ensure that the alignment step I'm currently using is suitable for this purpose. Below is the STAR command I'm using:

STAR --runThreadN 8 --genomeDir  --readFilesIn sample_1 sample_2 --sjdbGTFfile gff3_file --outFileNamePrefix --outSAMattributes All --outSAMtype BAM SortedByCoordinate --outSJfilterReads All --outFilterIntronMotifs RemoveNoncanonical --alignSJoverhangMin 8

Would you recommend any changes to make this pipeline more suitable for MAJIQ?

Thanks,

Rotem 

[San Jewell]

Hi Rotem,

I think that most of this alignment command looks correct, however, the part "--outFilterIntronMotifs RemoveNoncanonical" seems by the STAR manual to drop alignments which aren't canonical. This would likely remove a large amount of data towards detecting denovo events in majiq, if that's one of the things you'd like to use majiq for. Majiq requires de novo junctions to be represented by multiple reads at multiple positions so we sort of implicitly remove sjs that are one-offs in the STAR alignments. Depending on your use case I'd recommend dropping that one argument.

Thanks,

-San