RNASeq or WGS for fusion search?

[Jane M]

Dear all,

I am interesting in characterising tumor samples from a few patients.
We will perform WES for sure.
We hypothesised that there could be fusions as driver events. Thus, we are wondering whether RNASeq or WGS would be the best for this purpose.

Here are the pros and cons that I see in our case:

RNASeq:
+: high coverage, expressed fusions
-: impossible to detect a fusion that leads to gene extinction (we won't have normal cells to compare with), not sure to have high quality RNA, if our hypothesis is wrong (no fusion), we cannot look for more events.

WGS:
+: less quality issue for sequencing, all events expressed or not, detection of somatic variants in non coding regions like promoters
-: low coverage (maybe ~10X because of price), repeated regions are hard to analyse, need a control (-> price*2)

Do you have recommendation in this context?
Thank you in advance,
Jane

[Bukowski]

Picking it out of WGS data is hard, because read support for any event is low. Personally I've had great results from RNA-Seq data in cancer - picking out new and characterised fusions with FusionCatcher. However, obviously relies on the expression of that fusion event as you point out.

[Jane M]

Thank you Bukowski for your answer.
Which coverage is needed to detect fusions in your opinion in WGS data? Do you think paired control cells are needed?

I though using CRAC and tophat --fusion-search. What are the advantages of FusionCatcher? Do you have an idea of the false positive rate?