unmapped short and RSEM usage

[Varun Gupta]

HI Alex
Hope you are doing well. I am trying to map a dataset with STAR, but I am getting almost 99 % reads as Unmapped short


Time    Speed        Read     Read   Mapped   Mapped   Mapped   Mapped Unmapped Unmapped Unmapped Unmapped
                    M/hr      number   length   unique   length   MMrate    multi   multi+       MM    short    other
Jan 26 20:29:30    792.1    13200902       50     0.7%     37.9     1.3%     0.0%     0.2%     0.0%    99.1%     0.0%
Jan 26 20:30:30    813.4    27113245       50     0.7%     38.0     1.3%     0.0%     0.2%     0.0%    99.1%     0.0%
Jan 26 20:31:30    823.4    41170736       50     0.7%     38.0     1.3%     0.0%     0.2%     0.0%    99.1%     0.0%
Jan 26 20:32:30    832.8    55518054       50     0.7%     38.0     1.3%     0.0%     0.2%     0.0%    99.1%     0.0%
Jan 26 20:33:30    836.6    69720116       50     0.7%     38.0     1.3%     0.0%     0.2%     0.0%    99.1%     0.0%
ALL DONE!

Of course the reads are 50bp and also have adapter and primers. Now STAR being a local aligner, can deal with adapter sequences, but what parameters should I change to have a good mapping.

Should I change these options to 0

--outFilterScoreMinOverLread  --outFilterMatchNminOverLread

Also since I want to use RSEM to calculate isoform expression/quantification, does STAR for the time being onlyoutputs perfect matches(no soft clipping) in Aligned.toTranscriptome.out.bam files? This is because I cannot see any alignment with Soft clip in the cigar string.

Regards
Varun

[Alexander Dobin]

Hi Varun,

if the poor mappability is caused by long untrimmed adapters, reducing --outFilterScoreMinOverLread  --outFilterMatchNminOverLread should help.

You could make them 0, but then you will get a lot of really short multimappers. I would  make them ~0.3, so the min mapped length is ~15.

Cheers

Alex