Newbler - Singletons and duplicates

[dschika]

Hey,

Analyzing the reads Newbler marked as singletons I found some duplicate reads. I'm wondering now if duplicate reads are marked as singleton/repeat/outlier in the ReadStatus file? Or in other words (please correct me if I'm wrong):
If Newbler finds a duplicate, it just uses one of those reads (as long as one doesn't use -ud option), but how are the other read(s) marked?

As I want to construct an unigene set I simply wanted to take all the reads marked as singletons. Using the TrimStatus-file I constructed a file with the trimmed singleton sequences. After removing the duplicate reads is there anything else I have to take care of before taking the singletons as singletons?

Regards,
Thomas

[flxlex]

Duplicate reads (the result of two beads in one microreactor during emPCR) are all used by newbler and can therefore end up as singletons, (partially) assembled, repeat etc. However, during base and quality consensus calling, and for the calculations of read depth, only one of the copies (presumably the 'best' one) is used. Except, as you write, when the -ud flag is set.

So, newbler marks the duplicates the same way (I assume, I have not checked this...).

And as far as I know, you are correct in that after duplicate removal, you have your set of singletons.

One thing though, you could try to map the singletons to the assembly and see if some are actually aligning after all. We noticed that when we used the -large option, a significant portion of the singletons were in fact mappable to the assembly. The explanation we got from 454 for this had to do with how -large assemblies are more stringent during alignment (to speed up that phase), thereby kicking out more reads incorrectly as singeltons.

[dschika]

Thanks for your reply, flxlex!
Forgot to mention that I'm doing a denovo transcriptome assembly without the -large flag.
Nevertheless I mapped the singletons to the assembly using ssaha2. 3.5% mapped perfectly to the isotigs. Not a very large number, but at least the number of singletons goes down a bit