Clarification for 2pass per sample vs multi sample for transcriptome

[Patrick Tran Van]

Hi,

My goal is to assemble a transcriptome with trinity.

I have 3 RNAseq samples, I am a bit confuse about the right way to do, is it valid to :

1) Perform per-sample 2-pass mapping with the option --twopassMode Basic as described on p15 of the manual (so 3 times because I have 3 samples)

2) Merge the 3 .bam files and feed it to trinity.

Thanks for your help !

[Alexander Dobin]

Hi Patrick,

if you want to get just one BAM file merged from the 3 samples, the easiest (and most accurate) thing to do is to feed all three samples together in one STAR job with the --twopassMode Basic option for 2-pass. The the junctions detected in the 1st pass from all samples will be used

For SE reads - comma separated list:

--readFilesIn sample1.fq,sample2.fq,sample3.fq

For PE reads - comma separated list for read1 <space> comma separated list for read2

--readFilesIn sample1_read1.fq,sample2_read1.fq,sample3_read1.fq      sample1_read2.fq,sample2_read2.fq,sample3_read2.fq

Cheers

Alex

[Patrick Tran Van]

Thanks Alexander !

I have a mix of paired end and single end, can I mix them ?

[Alexander Dobin]

Hi Patrick,

single- and paired-end reads cannot be mapped in one pass - you have to run them separately.

You can merge the resulting BAM files.

Cheers

Alex

[Patrick Tran Van]

Thanks,

Quick question, if I have single reads that are reverse stranded and PE that are RF stranded, do I need to specify some options ?

[Alexander Dobin]

Hi Patrick,

I believe SE reverse are directly compatible with RF (1-reverse, 2-forward), so you can merge the BAM files without losing strand information.

Of course, this depends on the downstream software - it needs to know these conventions.

Cheers

Alex

[Patrick Tran Van]

No but I mean, STAR does not make a distinction between FR, RF, F or R right ? there is no default option for orientation ?

[Alexander Dobin]

Hi Patrick,

STAR does not use the library strandedness information. It maps the read sequences as they are in the FASTQ file and reports the DNA strands they map to.

Cheers

Alex