samtools XS:A optional field

[Bioinyang]

Early RNA-seq protocols were unstranded ie the RNA may have originated on the plus strand, but because the library was created from double stranded cDNA you obtain reads from both the plus and minus strand - in this case the samtools flag for which strand the read mapped to doesnt tell you which strand the RNA it originated from. However, if the read is across a splice junction you can work it out (as charlie_sequencing posted) hence the tophat XS flag. Reads from an unstranded library which do not cross a splice cannot be assigned a strand.

However, lots of people are doing stranded RNA-seq these days where the reads only come from one strand which is a lot nicer.

Hi,frozenlyse, As you said, unstranded library we don't konw reads come from which strand. Now I have a dataset of dUTP and I use --library firststrand .Tophat said,"it is assumed that only the strand generated during first strand synthesis is sequenced.",that means, the reads come from minus strand. But I found a half of reads was taged "XS:A:+"

[flobpf]

Other posts that may be of help

Please check out these posts for a discussion of directional sequencing and XS:A tags:

http://seqanswers.com/forums/showthread.php?t=9303

and

http://seqanswers.com/forums/showthread.php?p=37298

[frozenlyse]

Early RNA-seq protocols were unstranded ie the RNA may have originated on the plus strand, but because the library was created from double stranded cDNA you obtain reads from both the plus and minus strand - in this case the samtools flag for which strand the read mapped to doesnt tell you which strand the RNA it originated from. However, if the read is across a splice junction you can work it out (as charlie_sequencing posted) hence the tophat XS flag. Reads from an unstranded library which do not cross a splice cannot be assigned a strand.

However, lots of people are doing stranded RNA-seq these days where the reads only come from one strand which is a lot nicer.

[charlie_sequencing]

Thanks so much for your reply, nilshomer and dagarfield. I found the Cufflinks manual http://cufflinks.cbcb.umd.edu/manual.html where XS field was specified. When a read is mapped to the ref across splicing junctions, the boundary of intron (GT-AG, or CT-AC) will tell you the value of XS: "+" for GT-AG and "-" for CT-AC. It tells you "which strand the RNA that produced this read came from".

This led me to think about another question: is TopHat able to report the strand of the read if it is completely contained within an exon, i.e., without N in the CIGAR string?

[dagarfield]

There are two places in a SAM file that, on the surface, *seem* to provide equivalent information -- the second column (0 vs 16, or something) and the XS:A field.

I have a vague memory that the first tells you about mapping relative to the reference sequence. The second (XS:A) tells you something about the direction/strand of the *read itself*.

I wish I could find the post I read that told me that. I first came across the distinction while trying to understand the input fields required for TopHat. You might give a try searching for "TopHat" and "XS field" together. Please post back if you figure it out.

Best of luck,

DG

[nilshomer]

Does anyone know what XS:A field in samtools output mean by any chance?

XS:A (A is the SAM attribute type) can take the values of "+" or "-". I thought "+" or "-" might be related to how the read is mapped to the ref (forward or reverse complement), but it turned out not the case, because forward or reverse complement read can take both values.

Thanks!

The optional tags are specific to the aligner that generated them. What aligner did you use? Take a look at the manual from that aligner.

[charlie_sequencing]

Does anyone know what XS:A field in samtools output mean by any chance?

XS:A (A is the SAM attribute type) can take the values of "+" or "-". I thought "+" or "-" might be related to how the read is mapped to the ref (forward or reverse complement), but it turned out not the case, because forward or reverse complement read can take both values.

Thanks!