The first filter is to get all the variants that are in COSMIC but would be filtered because that are in a public database (gnomAD, TOPMED). So there is not reason to have the DB_exclude
In the second filter we grab any variant that is not in the public databases, novel likely somatic events and rare inherited variants
Then these are merged to get the likely somatic candidates.
https://github.com/tgen/phoenix/blob/9f04da2fdcce834e152158cf437ba7be8a9969d3/utilities/variant_filtering.jst#L44
This exclude line is not needed I believe:
The first filter is to get all the variants that are in COSMIC but would be filtered because that are in a public database (gnomAD, TOPMED). So there is not reason to have the DB_exclude
In the second filter we grab any variant that is not in the public databases, novel likely somatic events and rare inherited variants
Then these are merged to get the likely somatic candidates.