New isoform discovery with paired end reads across different exons

Question

[thomasterTW]

I am trying to figure out which isoform of a protein is expressed in my cells. I have reads in exons suggesting that either multiple isoforms are expressed, or a yet undiscovered isoform is expressed. To test this I need a tool that uses paired end reads across different exons in order to 'link' those exons together, so that I know that these exons are expressed as one isoform.

To elaborate on the situation:

Isoform1: |---[exon1]-----------------[exon3]---|

Isoform2: |---[exon1]---[exon2]-----------------|

my track: |---[exon1]---[exon2]---[exon3]---|

So there is a chance that isoform 1 and 2 are both expressed, or that my track is actually a new isoform. If I have reads with one end in exon 1 and the other end in exon 2, I know that isoform 2 is expressed. If one end is in exon 2 and the other in exon 3 I know that a yet unknown isoform is expressed.

Can someone suggest a tool to do this type of analysis?

Answer 1

[colindaven]

Cufflinks is a bit obvious so you have probably tried it already.

I would recommend a full de novo analysis using eg Trinity / CLC and map back to the genome with GMAP. Then you can see any new isoforms quite easily.

Answer 2

[EagleEye]

A: What is best tool to call diferentially expressed isforms ?