GeneFuse ignores fusions in TCF3

[nvolkovaGEL]

Hello,
Thanks for developing this tool! I was testing it on some hematological samples with TCF3 gene fusions (confirmed by FISH), and it seems like GeneFuse is not picking them up. I have checked that the gene is correctly reflected in the reference, and the reads connecting TCF3 and its partners are uniquely mapped to its location in the corresponding BAM files. I have two samples with different fusions, and in both of them, the fusion was not predicted even as untranscribed. Can you suggest what else can I try to fix this problem?
Thanks in advance!
Cheers,
Nadezda

[sfchen]

What's the fusion partner gene of this TCF3 fusion?

Please note that, GeneFuse only detects gene fusions that both two fusion genes are listed in the CSV file.

[nvolkovaGEL]

Hi, the fusion partners were HLF in one sample and PBX1 in the other, these fusions were experimentally confirmed by FISH. Both HLF and PBX1 were included in the fusion file along with TCF3, with the transcripts covering expected fusion junctions:
grep HLF test_fusions.csv

HLF_NM_002126,chr17:55264959-55325187

grep PBX1 test_fusions.csv

PBX1_NM_002585,chr1:164559183-164851831

grep TCF3 test_fusions.csv

TCF3_NM_003200,chr19:1609291-1652615

And these are the expected breakpoints:
sample 1:

chr19:1617928 - chr1:164683443, 59 supporting reads, TCF3-PBX1

sample 2:

chr19:1617921 - chr17:55318881, 77 supporting reads, TCF3-HLF
chr19:1617932 - chr17:55318898, 66 supporting reads, TCF3-HLF

Is there any other reason why this may be happening? Any hard coded filters on the reads/regions? Many thanks in advance!

[renyongzhe]

Maybe reads contained breakpoint were filtered. In such a scenario which a mismatch occurred beside the breakpoint, all kmer contained the mismatch and the breakpoint would not be traced back to a genomic position and thus caused more mismatch and be filtered eventually.