Structural variation detection from RNA-seq

Question

[Hans]

Hello I have two sets of reads that were obtained from RNA bulks of resistant and susceptible wheat recombinant inbred lines. I am looking for polymorphism between the bulks. Based on previous results, I suspect that the polymorphism associated with the resistance is a large insert of 1-5kb (up to the whole gene). I also suspect that this insert is not present in the reference genome. I am looking for a software that can detect large structural variations using de-novo assembly.
Can you please recommend which software to use? Thank you

[WouterDeCoster]

I don't think RNA-seq is the most appropriate technology for this, but now that the data is already generated we better make the best out of this.

I'll assume you don't have a reference genome since you want to go for de novo assembly?

Based on previous results, I suspect that the polymorphism associated with the resistance is a large insert of 1-5kb (up to the whole gene).

So you expect an insertion of an entire new gene? That gene should be expressed before you will see anything in the transcriptome, right?

[Hans]

Thank you for the reply. I expect an insertion of an entire new gene, I know that resistance genes have in many cases presence/absence variation. I expect that the gene will be expressed because the resistance was active at the time that the samples were taken. However, the expression levels may be low. I do have a refernce genome and I have conducted SNP and short indel discovery with this data. This got us closer to the target but looking at the refernce genome we found no candidate genes.

[WouterDeCoster]

An entire gene wouldn't map to the reference genome, I would try a de novo assembly on the unmapped reads.

Note that I haven't done anything like this before - it's my first intuition ;)

[Hans]

Yes, this is what I am looking for, de-novo assembly and comparison of long sequences.

[WouterDeCoster]

For RNA-seq the most commonly used de novo assembler is trinity

[Friederike]

what do you mean with "looking at the reference genome we found no candidate genes"? did you see a large insertion or not? and why do you hope to see more relevant information from the transcriptome data?

Answer 1

[colindaven]

Interesting project. I would approach it like this:

• Trimming - very important pre de novo assembly
• de novo assembly eg with trinity or Soap denovo trans or CLC or Abyss
• extract ORFs with transdecoder
• functional analysis of transcripts with interpro scan.
• mapping to the genome with gmap
• check mapped and unmapped and partially mapped genes in the area of interest - if known
• orthology mapping - eg with proteinortho - for transcripts unique to one set. Join this information with the IPS scan info

[Hans]

Thank you for the advise, I will give it a try.