Assembly Of Unmapped Reads

Question

[learnerforever]

From our recent HiSeq runs, we are taking unmapped reads from BAMs and assembling. Can someone recommend a good workflow to get the most out of unmapped reads? Current plan is:

1. Assemble using Abyss (or some paired-end assembly tool)
2. Get contigs more than 200bp (randomly chosen threshold--open to ideas) and blast against hg18
3. Take unmatched contigs and blast to see non-hg hits

Any ideas/suggestions welcome1

Thanks, forum!

Answer 1

[Zhidkov]

Actually what you suggesting is the approach used in Human Microbiome project (so you can look there for a pipeline, how to exclude from your data "human contamination"). If during mapping to human genome you did not demand for "mapped unique only", then I think you can skip the step of blast contigs from de novo assembly against hg and blast them against non human sequences.

Ilia

Answer 2

[Istvan Albert]

First I would say you would need define what are you hoping to achieve with the process:

1. are you trying to identify the contamination of your samples?
2. are you trying to recover reads that may have been mistakenly left unmapped?
3. are you trying to identify some systematic errors in your sequencer/sample preparation?

Depending on your sample preparation you may have various contaminants that may not be interesting at all.

[learnerforever]

Current goal is to check for "non-human" DNA in the sample, with the hope that reads that did not align against hg18 might have a homology with something else ...