Scaffolding on Trinity output

[vantolam]

I have done a de novo transcriptome assembly with Trinity using pair end reads,but when I mapped the reads back to the contigs I often observed that one of the two pair reads mapped back to a contig and other read mapped back other conting. I would like to join the obtained contigs into scaffolds using the information of the pair end reads. What programs can I use to do that?

[nima.rafati]

Hej

It might be a late response but I share my experience. Let me begin by saying that new version of Trinity does the scaffolding but I am not sure in which step.
Regarding scaffolding among available tools I have used SSPACE from base clear company but it did not really help, it might work for your data.
Considering setting the same component/contig name for isoforms, alternative splices or different variants by Trinity, I used CAP3 and extracted built contigs belonging to the same component.

Hope it gives you some idea. Please let me know if you have found another way. Or if you can also help me to improve.

Cheers,
Nima

[gringer]

Scaffolding should already be done by Butterfly in the Trinity process. However, I don't think it will combine paths unless there is good support for that combination (i.e. more than a couple of read pairs).

[christinawu2008]

Scaffolding should already be done by Butterfly in the Trinity process. However, I don't think it will combine paths unless there is good support for that combination (i.e. more than a couple of read pairs).

Hi there,

I think so, but there is no 'N's in the output assembly, which were generated from SOAP-trans.. which makes me think Trinity discards the transcripts containing Ns after doing scaffolding and gap filling? Does Trinity do gap filling?

Thanks

Christina

[gringer]

I posted this question to the Trinity users mailing list, and got the following response:

Trinity uses the pairing info, but doesn't (yet) join contigs across sequencing gaps. This is something that could be engineered into Butterfly (not so easy), or added as a post-process that would run on the per-component level (much easier - could be fairly straightforward). If there's good pairing to support the scaffolding, then the relevant contigs should be grouped together into a single Chrysalis component - and all the reads you'd need to do the scaffolding are right there.

It's been on the to-do list for a very long while, but just hasn't risen to a high priority. It'll happen eventually. It's mostly an engineering task.

[christinawu2008]

I posted this question to the Trinity users mailing list, and got the following response:

Trinity uses the pairing info, but doesn't (yet) join contigs across sequencing gaps. This is something that could be engineered into Butterfly (not so easy), or added as a post-process that would run on the per-component level (much easier - could be fairly straightforward). If there's good pairing to support the scaffolding, then the relevant contigs should be grouped together into a single Chrysalis component - and all the reads you'd need to do the scaffolding are right there.

It's been on the to-do list for a very long while, but just hasn't risen to a high priority. It'll happen eventually. It's mostly an engineering task.

Thanks gringer

So seems to need some additional scaffolding steps from other programs...

Regards,
Chen