Is there a way to use transitivity of homology directly in the PISA/DISTAL? Or should we always leave that to the subsequent clustering step.