I currently have a feature driven walker in gatk-protected that is slower than a read-walker equivalent due to the repetitive Read iterator re-instantiation when accessing overlapping reads using the ReadContext.
Ideally the engine (FeatureManager?) would try to reuse open read iterators instead of creating them for each feature/interval.
I currently have a feature driven walker in gatk-protected that is slower than a read-walker equivalent due to the repetitive Read iterator re-instantiation when accessing overlapping reads using the ReadContext.
Ideally the engine (FeatureManager?) would try to reuse open read iterators instead of creating them for each feature/interval.