How to extract assembled transcript sequence from RNA-seq data instead of ref genome?

[sindrle]

Lets say you have called indels and SNPs with GATK. Would that work, or can you please share some more details?

I have never done this before.

[Jeremy]

It would probably not be too difficult to get a list of variants between your sample and the reference, then convert the reference genome to the variant bases and then use gtf to fasta to get the variant transcripts. I have done something similar in R using the seqinr package.

[lzu]

sorry, can't find one from the top of my head but the 'normal' mapping procedure after de novo assembly of transcripts should do the job pretty well. Just account for your diversity of strains when you choose the mapping parameters. Use loose mapping criteria after your assembly and it should be fine. If, however, this doesn't give you the desired results, what I normally do is to BLAST the transcripts against an in-house database. This is even looser than what most of the mappers allow . Also, if the transcripts are becoming too long this should be the way to go.

Hope that helps:


FWIW: see below some papers for assembly and mapping which might be helpful anyways.

There you go:
Garber et al.
Trinity used to assembly transcripts
Oases assembler

----
Thanks for the suggestion. I read some papers which use model reference genome to predict alternative splicing diversity of subspecies or species 'variety' with RNA-seq data. There might be errors in results if some exons or introns are truly physically lost in those subspecies/variety genome due to the genetic diversity among different groups/populations...

[sphil]

sorry, can't find one from the top of my head but the 'normal' mapping procedure after de novo assembly of transcripts should do the job pretty well. Just account for your diversity of strains when you choose the mapping parameters. Use loose mapping criteria after your assembly and it should be fine. If, however, this doesn't give you the desired results, what I normally do is to BLAST the transcripts against an in-house database. This is even looser than what most of the mappers allow . Also, if the transcripts are becoming too long this should be the way to go.

Hope that helps:


FWIW: see below some papers for assembly and mapping which might be helpful anyways.

There you go:
Garber et al.
Trinity used to assembly transcripts
Oases assembler

[lzu]

Yep, maybe that's the better way of doing it. Assemble the transcripts denovo an map those transcript to the reference genome.

Do you know any paper(s) that "first denovo assemble RNA-seq, then map to ref genome"?

[sphil]

You are right, you've got my point. I still don't know how to extract sequences from the grape 'variety' RNA-seq data. Maybe it is hard, or should I assemble RNA-seq de novo by using Trinity?

Yep, maybe that's the better way of doing it. Assemble the transcripts denovo an map those transcript to the reference genome.

[lzu]

You are right, you've got my point. I still don't know how to extract sequences from the grape 'variety' RNA-seq data. Maybe it is hard, or should I assemble RNA-seq de novo by using Trinity?

[lzu]

Then I will end up getting a fasta file of transcript sequence extracted from the reference genome which is not what I wanted. I want the transcript sequences from my sample (a grape 'variety').

[sphil]

Cufflinks outputs a GTF assembly that annotates each of the loci that it calls. If you want a multi-fasta file of that, just use gtf_to_fasta (that probably comes with tophat, but if not you can google around for it).

be aware of where cufflinks gets the sequences from. maybe it uses the provided fasta and just extracts the sequecnes according to the coordinates given by the gtf. Which means, you end up with your 'non-variety' grape sequences.

[dpryan]

Cufflinks outputs a GTF assembly that annotates each of the loci that it calls. If you want a multi-fasta file of that, just use gtf_to_fasta (that probably comes with tophat, but if not you can google around for it).

[lzu]

Hi All,

I have a RNA-seq data from a 'subspeices' or 'variety' of grape. Grape genome is available.

I want to get the transcript fasta file of this 'species variety' after I mapped the grape 'variety' RNA-seq reads to grape reference genome via 'TopHat and Cufflinks' pipeline.

Cufflinks only produced output of 'transcript coordinate' file (positions of transcripts in the grape reference genome). But I need to extract the transcript assembly fasta sequence from this grape 'variety' RNA-seq data, not from the reference grape genome because there is a little bit evolutionary difference between my sample and the reference genome which I want to analyse later.

So how do I extract transcript fasta file from RNA-seq data of my sample instead of the reference genome after I ran the 'TopHat and Cufflinks' pipeline?

Thanks for your help and suggestion!

lzu