Impossible gene fusion annotations

Question

[Getting there]

Hello everyone,

I have some Bionano data which says that a deletion is causing a PAX5::ZCCHC7 gene fusion. We have some RNASeq data processed with STAR fusion data saying there is 1921 supporting reads for this gene fusion. However, how can a deletion cause this fusion when the two genes are on opposite strands? An inversion could cause this, but a deletion causing this makes no sense. Am I missing something here?

[Pierre Lindenbaum]

your previous questions have answers that should be validated: Noisy germline CNV data using CNVKit ; Does the GENOtation web-based PCA tool work? ;

Answer 1

[cmdcolin]

the strongest evidence for it being a "true gene fusion" would be probably be RNA-seq data with spliced reads joining the exons of both genes. you can try something like FusionInspector which shows the data on "fusion contigs" https://github.com/FusionInspector/FusionInspector. and indeed, like you say, if the RNA-seq data is truly joining the two genes, the might be an inversion type SV since it would be very unusual for a gene product to be formed from opposite strands.

schematic of fusion inspector from wiki

result of fusion inspector, showing 'sashimi arcs' between the two gene models, indicative of a fusion

https://github.com/FusionInspector/FusionInspector/wiki