Gene fusion detection

[kenosaki]

Hi,
I am wondering if detection of fusion gene is possible using a GA's single but relatively long (70-100 bp) read RNA-Seq data.
I know some software can detect fusion gene from paired-end dataset, but not sure how it's possible in a single read.
Any suggested tools or method?

Thanks,
Ken

[krobison]

It is certainly doable but your sensitivity will be inferior to paired end data.

Bowtie can be "tricked" into just aligning one end or the other of your reads (using the 3' & 5' trimming options); these alignments could be post-processed to look for split reads.

Alternatively, you could split the reads & create two files as though you had paired-end data.

[kenosaki]

Alternatively, you could split the reads & create two files as though you had paired-end data.

That's what I thought at first, but I'm not sure where I should split the read out. As you know, the first half has higher quality than the last half has.

But thanks, bowtie could be my option.

[HESmith]

Hi Ken,

We used split-end alignment to map insertion elements, but the same approach can be used to identify any novel junctions (such as gene fusions). The most important consideration is the size of the gap between the ends, since that's where the novel junction maps. For 70bp reads, mapping 25bp from each end leaves a 20bp window where the junction can fall. A detailed description of this method was described in the February 2011 issue of Biotechniques.

-Harold

[JohnK]

Hi Ken,

We used split-end alignment to map insertion elements, but the same approach can be used to identify any novel junctions (such as gene fusions). The most important consideration is the size of the gap between the ends, since that's where the novel junction maps. For 70bp reads, mapping 25bp from each end leaves a 20bp window where the junction can fall. A detailed description of this method was described in the February 2011 issue of Biotechniques.

-Harold

can you cite this? i tried to find it but had no luck.

[HESmith]

The link is here. In the supplemental materials, we describe the identification of two large deletions by split-end reads that were further apart than the read length. A similar strategy would identify fusion genes.

Harold

[JohnK]

The link is here. In the supplemental materials, we describe the identification of two large deletions by split-end reads that were further apart than the read length. A similar strategy would identify fusion genes.

Harold

Thank you, Harold!

[kenosaki]

Thanks guys, I'll try them.