De novo assembly of transcripts originating from specific target regions

[rboettcher]

Hello everyone,

After alignment of my paired-end RNAseq data, I have isolated some regions of interest and extracted reads mapping to these specific "target regions". Now, my intention is to perform de novo assembly with these reads to analyse the possible transcripts originating from my target regions. Hence, I am searching for assemblers that make use of my BAM files to recreate the transcripts.

So my questions is simple: could you recommend some tools?

Please note that I am solely interested in the transcripts and their exon boundaries / junctions, so I do not care for expression levels / transcript abundance.

My first idea was to use cufflinks with its GTF-guide option, however, I read this option still has severe problems so if anyone has alternatives I'd be happy to give them a try.

Kind regards

[Torst]

Have you tried using your extracted read sets with de novo transcript assemblers like Trinity and Oases?

[rboettcher]

Hello Torst,

No, I have not tried Oases or Trinity yet. Can you give any recommendations which one is preferable and how they perform in comparison with cufflinks + GTF-guide?

Please note that my current workstation is limited to 12 cores and 60 GB of RAM.

Kind regards

[Mark]

It's not clear from your post whether de novo or mapping assembly is best for you. If the organism you are worrking with is well characterized in the regions of interest, then tophat and cufflinks (mapping assembly) should do a good job for you. If it is not, and the potential for substantial differences in the underlying genomic DNA exists, then relying on mapping assembly or use of a single read filter based on mapping to feed a de novo assembly may come up short.

[rboettcher]

Hi Mark,

Sorry, I forgot to mention that I am working with human data (hg19), this is why I considered cufflinks.

[Jan_R]

http://www.cgat.org/~andreas/documentation/pysam/usage.html#fetching-reads-mapped-to-a-region

Extracting reads from sam-files:
If you are familiar with python you might consider using the module pysam for that purpose. The link I posted shows how to do it.