Ancient DNA adaptor removal and read merging

[jimmybee]

Hi everyone,

I've seen a few people doing ancient DNA work on this forum and I thought they might be able to help me out on a Bioinformatics question related to the analysis of an ancient DNA sample sequenced using Illumina HiSeq. Ive been following a protocol written by Kircher et al. (http://www.ncbi.nlm.nih.gov/pubmed/22237537) regarding the merging of paired end sequence data for small insert libraries. I have a library insert size of between 180-300bp and I'm working with read lengths of 100bp. Is this protocol appropriate for this project?

Alternatively, is it just easier to remove adaptors, quality trim and do single-read mapping rather than merging?

Cheers

[Heisman]

I have not read that paper, but could you in more detail describe the wet lab steps, how many lanes of data you have, and what your reservations are regarding following what is done in that paper?

[jimmybee]

Its a protocols paper, so nothing biologically relevant. I was asking advice as whether the process of merging with small insert sizes is common, appropriate given the stats I have given in my post, and whether this was a standard for ancient DNA analysis. There's not many good Bioinformatics methods papers for ancient DNA out there so it would be great to get advice on what others use.

Only have one lane - about 158 million pairs. Wet lab steps isn't my area sorry..

[Heisman]

Its a protocols paper, so nothing biologically relevant. I was asking advice as whether the process of merging with small insert sizes is common, appropriate given the stats I have given in my post, and whether this was a standard for ancient DNA analysis. There's not many good Bioinformatics methods papers for ancient DNA out there so it would be great to get advice on what others use.

Only have one lane - about 158 million pairs. Wet lab steps isn't my area sorry..

Oh, I see, you mean merging the two paired ends if they overlap to yield one single end read? I thought you meant merging multiple lanes of data. I'll look at that paper when I get to work and can access it but my gut instinct is that since some of your insert sizes are >200bp then there is no point in merging; I'd rather align all of the data identically. But, I'll glance at that paper in a couple of hours.

[Heisman]

So glancing through the paper I think it's probably fine to follow it. If I had more time I'd read it in detail so perhaps somebody else will chime in.

[jimmybee]

Thanks mate. Looks like a good way to go, especially considering my lack of experience in the QC of small insert libraries in paired end sequencing

[technical vault]

I actually wrote something that did something similar for my masters project, unfortunately it wasn't as intelligent as this in choosing the most likely overlap. Alas I don't have access to the final code (it's buried on a UCL fileserver which i no longer have access to). However, it was based on this: http://almlab.mit.edu/vibrioGenomes/SHERA_temp/ which might be worth a look.

[jimmybee]

Ok no problem, I'll have a look at it. All good ideas

[Bucky]

Check out pandaseq, works pretty well:

http://www.biomedcentral.com/1471-2105/13/31/abstract