Differentiate DNA-Seq / RNA-Seq and Exome sequencing data

[Khanjan]

Hi there,

If you have situation that you have files from three sequencing platforms ( DNA-Seq / RNA-Seq and exome sequencing ) and do not know which belongs to which, can you distinguish them in some way?

I have an idea, but wanted to check what you guys think?

- Map some reads from each technology to one chromosome ( to save time and computing power )

- The reads that map everywhere, exons + introns = DNA-Seq reads

- Reads that map to exons +- 50bp = Exome seq reads

- Reads which have very less or zero coverage in intronic regions = RNA-Seq reads.

What do you guys think? Is there any other way? ( Other than files sizes )

Thanks a ton,

[westerman]

Seems reasonable. If you have a known chromosome. :-)

[Khanjan]

Seems reasonable. If you have a known chromosome. :-)

What do you mean by a known chromosome?

[westerman]

Try working with, oh, say data from a Whale. :-)

In some ways, if you are making the thought experiment of "how do I differentiate between different unlabeled data sets" then you might as well make it even more interesting -- try to differentiate between data that is both unlabeled in platform and species.

Given data from a human sample, then, yes your approach seems reasonable.

[Khanjan]

How about if we make bed files for the alignments of exome and whole genome?

Would we be able to differentiate them using the UCSC browser ? I believe the tracks for the two should look very different.

What do you guys think ?

[ulz_peter]

You could better differentiate between Exome and RNA-seq reads by watching the mean insert size, since aligning RNA-seq reads that span exons would lead to long insert sizes and/or split-reads. If the RNA-seq data is stranded than it's even easier.

I guess the easiest method for visualization is IGV (http://www.broadinstitute.org/igv/)

So take a subset from those reads, align them and watch them in IGV. (the only problem you might run into is a very shallow coverage which doesn't give you a nice picture in every gene...)