I implemented an adjusted version of the LPC metric that completely removes pairs from separate phase blocks from the equation rather than counting them as correct. This seems to alleviate the problem of high LPC values for large length scales that I noted in https://github.com/UCSC-nanopore-cgl/margin/pull/1. I've decided that this behavior is better, so I've made it the default, although the prior behavior remains available as an option.
I implemented an adjusted version of the LPC metric that completely removes pairs from separate phase blocks from the equation rather than counting them as correct. This seems to alleviate the problem of high LPC values for large length scales that I noted in https://github.com/UCSC-nanopore-cgl/margin/pull/1. I've decided that this behavior is better, so I've made it the default, although the prior behavior remains available as an option.