Border scores generated by opexebo.analysis.borderscore are barely correlated with those produced by +scripts.ScreeningBytemplate from BNT.
A bug is entirely plausible, but the most likely answer is the difference in field detection techniques, since BS is evaluated based on individual fields rather than rateMap. Generate a new test set using the BNT method .+analyses.placeFieldAdaptive (this can be called within Screeningby modifying the BNT parameter file to change p.fieldDetection2D to adaptive)
Secondary note: the test data so far is not curated; it's a random mixture of cells. A curated set of cells that do include some strong border cells would be useful (also relevant for #5, #6)
Border scores generated by
opexebo.analysis.borderscore
are barely correlated with those produced by+scripts.ScreeningBytemplate
from BNT.A bug is entirely plausible, but the most likely answer is the difference in field detection techniques, since BS is evaluated based on individual fields rather than rateMap. Generate a new test set using the BNT method
.+analyses.placeFieldAdaptive
(this can be called within Screeningby modifying the BNT parameter file to change p.fieldDetection2D toadaptive
)