Hi.
I have many exome sequence data in my linux sever and I have found many variants such as SNPs,CNVs and many structural rearrangements.
Among them, I am especially interested in rearrangements because of its effect of tumor development.
I tried to find gene fusion by running software but it seems that lots of currently tools is based on RNA sequence not DNA sequence(Exome or whole genome sequence).
Moreover there are no gene fusion tools based on exome sequence. Of course I could find structural rearrangment using tools like SVDetect or BreakDancer but I am not sure whether these tools detect variants reliable.
Therefore my question is below.
Why there are no tools that identify gene fusion in exome data ?
And It's another question.
I have found the gene fusion tool "FusionMap" based on RNAs. I read paper and find FusionMap find gene fusion from either RNAs or gDNA. but I don't know what gDNA is. gDNA include whole-genome and exome-sequncing?
below is full sentence from FusionMap paper.
FusionMap can detect fusion events in both single- and paired-end datasets from either RNA-Seq or gDNA-Seq studies and characterize fusion junctions at base-pair resolution.
I am very appreciated if you answer my two questions.
Thank you :)
2 answers
To find gene fusions in exome sequencing data the breakpoint must fall within or very near the captured exons. This is more the exception than the rule. Thus, any wes-based fusion caller will have a very high false negative rate, which is the reason why there are not many tools around.
That's the whole point. You have to be incredibly lucky to find a gene fusion in exome sequencing data for that very reason. The breakpoint of the fusion has to be located in an exon, targeted UTR, or VERY close-by (within 20-50 nucleotides of a target usually).
Even if the breakpoint isn't near a targeted region, you can sometimes detect a function if it's unbalanced, with a copy number alteration or loss of heterozygosity on one side of the breakpoint but not the other. Within a gene with an unbalanced fusion, you'll see the CNA or LOH in exons on either the left or right side of the gene, but not across the entire gene.
Here's a study that used array CGH (instead of WES) to detect copy number breakpoints and nominate a few promising fusion candidates for further validation from a much larger initial pool (disclosure - I'm a middle author): http://www.nature.com/ncomms/2015/150527/ncomms8174/full/ncomms8174.html
I am assuming you are one of the authors, and granted I haven't read the Online Methods as I don't have electronic access to the full text from my institution, but I read through the methods and results via ResearchNet and this seems like a pretty special case. You had aCGH data to classify the tumours and then sequenced using as custom designed panel. While the panel design is similar to those in standard exome sequencing you had intronic sequences included, specifically of the introns where the break points normally lie. There is a whole figure showing reads and read-pairs spanning the fusion junction.
This isn't normally the sort of data you have in whole-exome sequencing. Calling copy number variation and detecting LOH in exome sequencing is challenging on its own because there are so many variables that impact sequencing coverage and capture. And then exome wide you would be running some sort of downstream analytics looking at all possible LOH/copy-gain regions of coverage and predicting all possible fusion genes....
The sheer number of false positive hits in that analysis would be extremely large.
Yes, we searched hundreds of archival aCGH profiles to look for copy number breakpoints in MET, then used targeted sequencing of that gene including the suspected introns to confirm a handful of real fusions. In the context of this question, copy number profiles inferred from exome sequencing would replace aCGH, and then an additional assay would be needed to confirm the suggestive and relevant cases -- targeted DNA sequencing of the suspected intron, targeted RNAseq, FISH, nanostrings, something else.
It's much more efficient and effective to just use RNAseq to begin with, but mangfu100 here already has the exome data and SV calls, and is looking for more signals to follow up on. CNV/LOH breakpoints are another weak but usable signal of unbalanced fusions that the other answers here didn't mention.
Log in to answer this question.
-
Copy number variation detection tools within exome sequencing
written by mangfu100
-
Is annovar suitable for structural rearrangements instead of SNPs or INDEL?
written by mangfu100
-
How to find structural variants
written by 897598644
-
Question about analyzing structural rearrangements with detailed annotation.
written by mangfu100
-
Is there any suitable annotation tool for structural rearrangements?
written by mangfu100
-
How to download DNA sample files in dbGap?
written by mangfu100
-
How to identify structural variants manually?
written by mangfu100
-
Question about sequence order using a software tool
written by mangfu100
-
Question about strand orientation.
written by mangfu100
-
What Are The 'Copy Number Detection' Tools Out There For Exome Capture Ngs Data.
written by Prateek
There are ways to detect, would read this paper
Thank you for you reply.
I already know above nature. it uses dRanger with Breakpointer but i can't use dranger. it is yet open to normal user.
so I need another way to find it.
and you suggest FusionMap.. but i think it is based on only RNA sequence.
I am not sure that i can use it with Exome sequencing.
Yes, Basically All these tools are designed for RNA-seq reads, Exome sequencing is outdated. I was also asking this question to myself that what would be wrong if you use FusionMap for exome sequencing, . Neverthless, the paper I shared above, they have found fusions from exome sequencing data. Probably That should help you.
"Exome sequencing is outdated"? A bold statement..
Outdated? Seriously?
TCGA would be pissed off.
Although RNA-Seq would improve gene-fusion finding nowadays Exome capture has evolve a lot. We use intron spanning capture in our custon-made exome and, in theory, we could also find gene-fusion in exome. It would be great a tool to find gene-fusion in exome.