Filtering detected fusions
Detect Fusion Genes from DNA applies only limited filtering to detected fusions. We therefore recommend performing post-filtering using Filter on Custom Criteria:
- Remove low-confidence fusions by using a maximum P-value and a minimum Z-score.
- Remove fusions with atypical Frequency: low-frequency fusions may still be detected in deeply sequenced data, while high-frequency fusions may be false positives if sample purity is low.
- Require a minimum 5' or 3' read coverage: fusions with insufficient coverage are more likely to be false positives.
- Require a maximum Gene distance: breakpoints located more than 50-100 Mb away from the nearest gene boundary are most likely structural variants without functional impact on nearby genes.
- Remove intragenic breakpoints that lead to Reversal, as these are most often false positives.
For targeted sequencing data, false positives can be further reduced using Filter Based on Overlap, by removing fusions where both breakpoints are located outside the targeted regions.