RNA-seq: different fragments from same transcript?

[mlacencio]

Hi everyone!

I got trapped in the following situation: I have some pairs of transcripts obtained by RNA-seq that were annotated with same terms, e.g. contigs A and B were annotated as 2-alkenal reductase (nadp(+)-dependent).

When I blasted them against genomes (blastn), refseq_rna (blastn) or refseq_protein (blastx), it seems that these two transcripts are fragments of a larger transcript (please see the attached files). Is this possible in a RNA-seq? Or maybe these transcripts are just isoforms?

Transcripts were assembled by a de novo technique since the organism of interest lacks a sequenced genome.

Best,

Marcio

[alezhe]

welcome to real world :-)
it's the common situation in transcriptomic studies of species without sequenced genome.

de-novo assemblers just output contigs of some confidence. there can be a lot of chimers as well.

as variant you can run the pool of your contigs through a contig assembler like CAP3.
as example see http://www.biomedcentral.com/1471-2164/12/227

[mlacencio]

Hi alezhe!

Can CAP3 (or another assembler) work in my case considering that these transcripts are non-overlapping? When I used "Blast 2 sequences" to align these transcripts, no significant similarity was found.

Best,

Marcio

[alezhe]

so if these two fragments has no overlapping then your strategy depends on your goal.
what do you want to get from this transcriptome assembly?

[mlacencio]

First, I want to find the differentially expressed transcripts between two different conditions. Second, I want to assign to these transcripts gene ontology terms -- using blast2go or Argot2 -- to subsequently perform a functional enrichment analysis.

Best,

Marcio

[sarvidsson]

~500 bases is a bit short to get highly confident numbers but you should check whether you have similar FPKM values for both fragments in the different samples - or given that they are of similar length, raw read counts would be OK to look at. If so, they may come from the same transcript.

If they have different orders of magnitude of expression, one of the following applies:
- two different genes
- two different splice forms
- the pieces come from the 5' and 3' ends of one transcript, and you have high levels of RNA degradation in your samples

[mlacencio]

Thank you for the answers!

I will follow your suggestions to try to solve this issue.

Still regarding RNA-seq, in this same set of contigs I got some pairs of contigs that were annotated with same terms, e.g. contigs A and B were annotated as 2-alkenal reductase (nadp(+)-dependent), and they ARE significantly similar according to blastn (see attached file).

Is the alignment in the attached file an indicative of assembly error? It seems that these two contigs could be merged in only one larger contig. Can anyone try to explain to me what is going on here? alezhe, maybe this situation could be solved according to your suggestion!

The assembler used was the CLC Genomics Workbench software. I don't know the parameters used because I just received the contigs for further analysis. Although I have never assembled RNA-seqs reads before, the annotation I performed suggested me that something was wrong.

Best,

Marcio

[sarvidsson]

Is the alignment in the attached file an indicative of assembly error? It seems that these two contigs could be merged in only one larger contig. Can anyone try to explain to me what is going on here?

It might be an assembly error, but to track that down you need to look at the complete dataset - you would need to look at the read counts of each transcontig and how many read pairs they "share". Without that data you can only speculate.

The assembler used was the CLC Genomics Workbench software. I don't know the parameters used because I just received the contigs for further analysis. Although I have never assembled RNA-seqs reads before, the annotation I performed suggested me that something was wrong.

The support at CLC isn't bad, you should ask your colleague to describe the issue to them - given enough data they should be able to explain why the assembler made that choice.
However, the CLC Genomics Workbench doesn't contain a transcriptome assembler - the de novo assembler is quite OK for certain tasks (although a bit of a black box and not very configurable), but doesn't seem to try to tackle RNA-Seq-specific problems (large coverage variation, multiple spliceforms...).