Skip to content

Latest commit

 

History

History
34 lines (23 loc) · 1.09 KB

score.md

File metadata and controls

34 lines (23 loc) · 1.09 KB

Scoring fusion

One disadvantage of the tools is that they tend to report false positive results. Not all fusion tools provide estimated score for a fusion. The only way how to correctly verify a fusion is biologically (RT-qPCR …).

fusion-report uses weighted approach of assigning weights to tools and databases. By default, each tool is assigned the same weight. This is because each tool uses different approach of discovering fusions and report different results, for example FusionCatcher will work best on somatic samples.

You can customize weight of individual tool by specific parameter <tool-name>_weight 30. The sum of the weights has to be 100!

Weights for databases are as follows:

  • COSMIC (50)
  • MITELMAN (50)
  • FusionGDB2 (0)

It is strongly suggested to use all supported databases in order to get the best estimated score.

Formula

The final formula for calculating score is:

$$ score = 0.8 * \sum_{tool}^{tools provided} f(fusion, tool) + 0.2 * \sum_{db}^{dbs provided} g(fusion, db)*w(db) $$

Weights for databases are as follows:

  • COSMIC (50)
  • MITELMAN (50)
  • FusionGDB2 (0)