RNASeq de novo from TopHat unmapped.bam

[demold]

I ran a RNASeq pipeline for human skin samples using TopHat2 and hg19 as the reference genome and, because I wanted to find contaminants from other organisms, I used the reads contained in unmapped.bam as input (after convert them to fastq files) for de novo RNASeq with Trinity.

I don't understand why some contigs generated by Trinity map with 100% identity in about 3000 bp alignment to ensembl transcripts.

Why the reads from which that contigs were generated were in unmmaped.bam file produced by TopHat?

Any idea?

[Zunaira]

Hi,

I read your post. I am also trying to construct a de novo assembly using unmapped.bam files from TopHat but I am getting errors. Did you sort your .bam reads before converting them to fastq? I know you did it long time ago but Can you please help me by sharing the steps?

[offspring]

Hi Zunaira,

you haven't stated the exact problems you see, but TopHat unmapped.bam files generally contain several errors that can impede processing.

We published a software called TopHat-Recondition to fix these errors a while ago, please see the paper and the software.

Let me know in case you have questions.