Table 1 Performance comparison of individual fusion calling algorithms. Fusion calling algorithms utilized by EnFusion and their contributions to fusion calling in the NCH pediatric cancer and hematologic disease cohort
From: Discovery of clinically relevant fusions in pediatric cancer
Tool | Version | Aligner | Reference | Average fusions called per case | Sensitivity (clinically relevant fusions called out of 67) |
|---|---|---|---|---|---|
Arriba | v1.2.0 | STAR aligner | Uhrig et al., 2021 [29] Genome Res | 54 | 88.1% (59) |
CICERO | v0.3.0 | candidate SV (structural variant) breakpoints and splice junction | Tian et al., 2020 [30] Genome Biol | 1909 | 92.5% (62) |
FusionMap | v mono-2.10.9 | GSNAP (Genomic Short-read Nucleotide Alignment Program) - 12mer based | Ge et al., 2011 Bioinformatics [31] | 34 | 86.6% (58) |
FusionCatcher | v0.99.7c | 4 aligners to identify junctions (Bowtie, BLAT, STAR, and Bowtie2) | Nicorici et al., 2014 [32] bioRxiv | 1554 | 89.6% (60) |
JAFFA | direct v1.09 | BLAT, uses kmers to selects reads that do not map to known transcripts | Davidson et al., 2015 [33] Genome Med | 1134 | 97.0% (65) |
MapSplice | v2.2.1 | approximate sequence alignment combined with a local search | Wang et al., 2010 [34] Nucleic Acids Res | 37 | 85.1% (57) |
STAR-Fusion | v1.6.0 | STAR aligner | Haas et al., 2019 [25] Genome Biol | 71 | 94.0% (63) |