Are any additional parameters for STAR to hand the stranded RNA-Seq data?

[Dapeng Wang]

Hi there,

If I plan on using STAR-featurecount-DESeq2 to analyze the stranded paired-end RNA-Seq datasets, I am wondering if there are some specific parameters needed by STAR to determine which type of the stranded library it is?

Alternatively, is STAR very smart to automatically determine the library type itself?

Many thanks.

Kind regards

Tom

[Alexander Dobin]

Hi Tom,

presently STAR does not use strand information for mapping, i.e. there is no filtering of reads that map opposite to the annotated strand of the gene.

If you use STAR's --quantMode GeneCounts option, it will output counts for 3 types of strandedness: unstranded, forward, reverse, so you would have to select the right orientation before sending it to DEseq. If you count with featureCounts, you would have to specify the option to count alignments in a certain strand orientation.

Cheers

Alex

[Dapeng Wang]

Hi Alex,

Thanks for the explanation.

1. For the stranded RNA-Seq data, I don't need to specify the strandness information for STAR. But I need to specify the strandness information for featureCount by using strandSpecific. Am I right?

2. For the unstranded RNA-Seq data, I know for Cufflinks/Cuffdiff downstream analysis, I should specify the argument --outSAMstrandField. But for featurecount downstream analysis, do I need to specify any specific argument for STAR during the mapping process?

3. If I need to feed the uniquely-mapped BAM file to featurecount, is there any parameter for STAR to only output the uniquely-mapped reads in a BAM file? Or it there any other way to get around this?

Many thanks,

Tom

[Alexander Dobin]

Hi Tom,

>>> 1. For the stranded RNA-Seq data, I don't need to specify the strandness information for STAR. But I need to specify the strandness information for featureCount by using strandSpecific. Am I right?

You are right, use -s 0/1/2 option from featureCounts.

>>> 2. For the unstranded RNA-Seq data, I know for Cufflinks/Cuffdiff downstream analysis, I should specify the argument --outSAMstrandField. But for featurecount downstream analysis, do I need to specify any specific argument for STAR during the mapping process?

 --outSAMstrandField add XA strand flag for spliced alignments only. AFAIK featureCounts does not use this flag.

>>> 3. If I need to feed the uniquely-mapped BAM file to featurecount, is there any parameter for STAR to only output the uniquely-mapped reads in a BAM file? Or it there any other way to get around this?

featureCounts uses NH tag value to detect the unique/multimappers, which STAR outputs by default, so you do not need to output unique only mappers in STAR.

Cheers

Alex