How should a strand-specific experiment look like in IGV?

[panos_ed]

I have a paired-end, strand-specific dataset and used Trinity to assemble the contigs. I then mapped the reads back to the contigs to calculate transcript abundance and viewed the resulting bam file in IGV (don't have a reference genome).

Is there a way that I can tell whether my dataset is really strand-specific just by looking at the picture in IGV? If I had single-end sequences I should expect to have reads only in one strand (i.e. pointing to the right). Am I right?

But since I have paired-end reads, what the IGV picture should look like?

[Jim Robinson]

Hi,

It depends on the library details, but you should could try "color by first-in-pair" from the popup menu. This will color by the strand of the template, rather than the read, at least for libraries I am familiar with. Also, igvtools has options to compute coverage by pair strand, although this isn't available in igv itself yet. Better strand support is on our radar.

Jim

[panos_ed]

Thanks Jim!

I selected "first-of-pair" as you told me but can't find out what the colors mean (and couldn't find it either in IGV's manual).

Can you please explain what the colors (red, blue and grey) mean?

[Jim Robinson]

red = neg strand (first read of the pair is no the neg strand)
blue = pos strand
grey = pair information not available, so "first of pair" is undefined

[panos_ed]

Thanks for the info Jim!

So if most of the pairs in a contig contain reads that are like this:

=====>----------<=====
(red)           (blue)

then this means that sequencing was strand-specific.

Am I right?

What if both of them are red or blue?

[Simon Anders]

No, even in a non-strand-specific library, the two reads of a pair should point towards each other (or away from each other if it was a library-prep protocol with circularization). You should check whether it is predominantly the first read that point in the sense direction of the gene it aligns to.

[Dario1984]

red = neg strand (first read of the pair is no the neg strand)
blue = pos strand
grey = pair information not available, so "first of pair" is undefined

If I have Illumina directional RNA-seq data, should the colours be the opposite or the same as the strand of the RefSeq gene, when colouring by the first read of the pair ? My colours are opposite to the RefSeq gene, but when isoforms are assembled by Cufflinks with --library-type fr-secondstrand, the isoforms are, as expected, on the same strand as the gene.

[stanikay]

So is directional RNA-seq the same as strand-specific RNA-seq? any clarification will be greatly appreciated.

Thanx

[panos_ed]

According to this, directional and strand-specific are the same.

[stanikay]

Thanx panos-ed..

[ArghavanAri]

Hi,

I am trying to align and count the reads of paired-end strand-specific RNA-seq data using Hisat and htseq-count.
This is how my IGV look like. https://www.dropbox.com/s/snpapn051b...ness.jpeg?dl=0
(Color alignment by "first-of-pair strand") (view as pairs).
I am confused now. Why all the read pairs are red for NFE2L2? and clur for HNRNPA3?
I know that my data is strand specific, and that this color is showing the first pair read. So why both pairs have the same color?
Also can anyone tell me what options I should use for Hisat (or Tophat2) and htseq-count?

Thanks,

[cliffbeall]

The two genes are on different strands, you can see by the little chevrons in the annotation track at bottom, so that is expected.

[gringer]

As others have mentioned, you need to consider the direction of mapping as well as the read number. For a standard Illumina strand-specific sample preparation, the reverse read is the first read:

• Reverse mapping, First read -- consistent with a gene model reading left to right on IGV
• Reverse mapping, Second read -- consistent with a gene model reading right to left on IGV
• Forward mapping, First read -- consistent with a gene model reading right to left on IGV
• Forward mapping, second read -- consistent with a gene model reading left to right on IGV

Just in case you weren't aware, Tablet (since v1.14.10.20) has a concordant / directional colouring mode that colours sequences appropriately.

[ArghavanAri]

Thanks for the information.
But I still cannot understand the difference between fr-firststrand & fr-secondstrand in Tophat2 (or RF & FR in Hisat).
Can anyone help me with that?

[cliffbeall]

Thanks for the information.
But I still cannot understand the difference between fr-firststrand & fr-secondstrand in Tophat2 (or RF & FR in Hisat).
Can anyone help me with that?

I am fairly sure fr-firststrand means read one (aka forward read or fr) corresponds to the reference genome sequence (firststrand). fr-secondstrand means read one corresponds to the reverse complement (secondstrand) of the reference genome. FR and RF would be a similar concept - but they are just making a shorthand where if the read one or forward read corresponds to the reference strand it maps to the left of the reverse read on a linear representation of the genome (FR) and if it corresponds to the reverse complement of the reference it maps to the right of the reverse (RF).

I hope that makes sense -- you might need to go back to DNA/genome structure and how Illumina works to really understand the above.