Transcript annotations in VCF are lost when data isn't found by FusionInspector

Description of the bug

We've noticed some issues with the latest version of rnafusion. Some lesser known fusions are only called by a single tool (e.g. arriba). FusionInspector is thus unable to detect the same fusion (because of it's reliance on Star-Fusion). This causes some INFO fields in the VCF file to be empty, while this data is available in the Arriba output files.

Example (censored data):

chr<?>	<?>	.	N	N[chr<?>:<?>[	.	PASS	SVTYPE=BND;CHRA=chr<?>;CHRB=chr<?>;GENEA=<?>;GENEB=<?>;POSA=<?>;POSB=<?>;ORIENTATION=nan,nan;FOUND_DB=nan;FOUND_IN=arriba;TOOL_HITS=1;SCORE=0.167;FRAME_STATUS=nan;TRANSCRIPT_ID_A=nan;TRANSCRIPT_ID_B=nan;TRANSCRIPT_VERSION_A=0;TRANSCRIPT_VERSION_B=0;HGNC_ID_A=<?>;HGNC_ID_B=<?>;EXON_NUMBER_A=0;EXON_NUMBER_B=0;ANNOTATIONS=nan	GT:DV:RV:FFPM	./1:0:0:0.0

Would it make sense in the cases where the data isn't found in FusionInspector to fetch the transcript annotations from the fusion call results themselves? I'll be happy to implement this but want to discuss this first before putting in the work.

Command used and terminal output

I sadly have not yet been able to reproduce this using public datasets

Relevant files

No response

System information

Rnafusion: 4.0.0 and dev

No relevant files or reproducible dataset are provided. Start by tracing how FusionInspector and Arriba fusion-call results populate VCF INFO fields, especially transcript annotations, then determine whether a fallback is defined. Done means annotations remain available for calls absent from FusionInspector and a reproducible test case covers the behavior.