RNAseq read alignments using STAR

[Maura Casey]

Dear Colleagues,
We are currently using the STAR aligner software (version 2.3.0.1) to align RNA sequences to a reference genome. The majority of inserts in our sequencing libraries have short fragment lengths (< 60bp) and consequently STAR returns an output file stating that we have a large number of reads that are too short to be mapped to unique regions of the reference genome. We would like to increase the number of mappable reads, but are unsure of the default read length threshold used by STAR and more importantly how we can lower this using the STAR command line. We were wondering if anyone could assist with these queries?

Best wishes,

Maura Casey

[GenoMax]

Maura: Look for posts by user "alexdobin" (STAR's author) on SeqAnswers to see if this has been answered before. You can search for those posts by using the "advanced search" under the "search" menu and then searching by username "alexdobin".

Alex is active on this forum so expect to get a direct answer as well.

[Maura Casey]

Hey thanks very much will do that.

Maura

[Chipper]

It is explained in the google group somewhere, you can try
--outFilterScoreMinOverLread 0.25 --outFilterMatchNminOverLread 0.25 to allow 50 bp alignments from 2x100 bp runs. Default is 0.66 which means alignments < 2*100*0.66 are discarded.

[alexdobin]

Hi Maura,

@Chipper's suggestion is right to the point and should allow you to map short pieces of you reads.

Another option would be to clip the adapter sequence that must appear on 3' of you reads if your fragments are so short. You can do it with an external of trimmer software, or use --clip3pAdapterSeq within STAR.

Cheers
Alex

[Maura Casey]

Hi Chipper,

That command worked for us thanks very much. Thanks Alex.

Best,

Maura.