how to understand gene fusion and how to call gene fusion ?

Question

[winter_li]

HI , I am thinking these three questions: 1. what's the detailed relationship between the breakpoint and gene fusion in WGS data,not RNA-seq data ?

1. How to judge the breakpoint via soft-clipped reads ?

   3, How to know which gene included at a breakpoint in the gene fusion ?\

I am new guy to gene fusion ,hope get your help！

thank you very much!

Answer 1

[Amitm]

hi, I can point you to an informative review published recently, here. In brief, 'breakpoint' is the physical event that might be leading to gene-fusion. Before NGS, breakpoint mapping would be at chromosome band level, but now you can get base-pair resolution. Soft-clipping is when only a portion of a contiguous stretch of (fastq) read aligns while the un-alignable portion might be from a different chromosome or same chromosome (but beyond the limit of paired-end distance or gapped alignment).

There are many tools available that can analyze/ predict such events. If you are just looking into DNA-sequence data (WGS), then structural variant callers can identify such events under the category of translocations. A recent comparison of callers is here. Most of these callers would do the annotation and tell you the gene pairs (under fusion), or would give a VCF file that can be annotated with tools like SnpEff.

[Amitm]

I must remark that for complex structural variants like translocations, I have not yet tested an automated annotation pipeline. SnpEff page says it has limited capabilities and Variant Effect Predictor doesn't list translocations as supported. But I am finding filtering out true calls from all the predictions (from any SV caller) is a bigger challenge. Just my experience..

[winter_li]

HI, thank you for your reply! The breakpoint mapping is the one of way to detect gene fusion,but I have another quesion now: if I got the location of two breakpoints, such as one point is chr1:172599726, another point is chr4:96089478, Next how to konw which gene happened at this kind of gene fusion ??

Thank you !

[Amitm]

hi, assuming that the caller hasn't given the gene information already, you can simply put those coordinates in UCSC genome browser or Ensembl, to know the gene those coordinates might be in. If you have more than a few coordinates to do such checks then I suggest that you download gene coordinates in BED file format and use bedtools to annotate. Alternatively, you could feed your coordinates to Ensembl BioMart, or UCSC Table Browser to get gene annotations.