How To Compare 2 Differential Expressed Transcripts From 2 Different De Novo Assembly?

Question

[nbvasani]

Hi Fellows,

I have generated 2 de novo transcriptome assembly using Velvet/oases and Trinity. Differential expressed transcripts was generated from both assembly using edgeR. Now I want to compare Differential expressed transcripts from both assemblies, to see how many transcripts are similar in both assembly. Hoping you guys can help me out.

I would really appreciate your feedback.

Thanks in advance.

Naresh

[Damian Kao]

So you've mapped the same datasets to two different assemblies? Are you trying to assess the quality of the assemblies somehow?

[nbvasani]

Yes, I have mapped same dataset to two different assemblies. No I am not trying to assess the quality of the assemblies. But at some point I would like to assess the quality of assemblies. Actually with one of the Differential expressed transcript list generated from velvet/oases, I already did manual annonation from NCBi website. So in order to save time and labor work, now I want to compare Differential expressed transcript list generated from trinity to list generated from velvet/Oases.

Thanks for your feedback.

Naresh

[Damian Kao]

I see. So you really just want to map the annotations from one assembly to the other assembly somehow?

[nbvasani]

Yes Damian Kao.

Answer 1

[Charles Warden]

Maybe you want to try something like these programs (which look for differences in read distributions first). They seem like a very reasonable strategy, although I have to admit that I've had some trouble getting NIKS to work (although it might have been an issue with my specific data and/or the fact that I didn't try super hard):

RUFUS: http://bioinformatics.bc.edu/marthlab/wiki/index.php/Andrew_Farrell#RUFUS

NIKS: http://sourceforge.net/projects/niks/

NIKS paper: http://www.nature.com/nbt/journal/v31/n4/full/nbt.2515.html

Otherwise, I think you would have to BLAST/BLAT your assembled contigs against a common reference and try to map the corresponding contigs (which won't be trivial)

[nbvasani]

Thanks Cwarden45 for your feedback.

Naresh

Answer 2

[nbvasani]

Hi Cwarden45,

I was able to solve this problem, with your help. First I blastn one of my DE gene list with reference annoted DE gene list. I got output from blastn in .xml format, which I opened in blast2go tool. In blast2go tool I was able to observe which gene were mapped to reference DE gene list.

Thank you so much.

Naresh

[Charles Warden]

Ok, thanks - I'm glad to hear that. My main concern was that the mapping wouldn't be one-to-one (especially with the Trinity transcripts, which can be pretty large). However, it looks like things worked out in your case!

[nbvasani]

Ya, you are right about large transcripts. In blast2go tool, it pick up first match out of 3 or 4 match based on evalue.

[pbigbig]

Hi nbvasani,

Currently I am doing some stuff similar to yours. I just want to know that if I performed de novo assembly of same data set using Velvet/Oases and Trinity, how can I check which assembly is the better one at the first glance ? Are they N50, maximum transcript length .etc. ? My aim at the moment is not yet going to do DE analysis, but to do a whole transcriptome assembly and annotation for reference of this new organism.

Thank you in advance for your suggestion!

Best regards,

Phuong.