So right now, we have a MinLength validator. However, there is one slight problem in that it requires us to load the trajectory to determine if the length is sufficient.
For FAH-size projects, this can add days to the analysis time.
I see several possible solutions:
(Current workaround): preprocess data by making a symlinked directory structure that lacks links for RUN/CLONE pairs with too few gens.
(Possible future solution): Add a min-gen option to ConvertData / Builder that loads an empty trajectory for RUN/CLONE pairs with too few gens.
The advantage of (2) is that it helps keep the entire FAH conversion pipeline contained in the Builder object. The disadvantage is that (2) isn't as elegant as our current Validator scheme.
So right now, we have a MinLength validator. However, there is one slight problem in that it requires us to load the trajectory to determine if the length is sufficient.
For FAH-size projects, this can add days to the analysis time.
I see several possible solutions:
The advantage of (2) is that it helps keep the entire FAH conversion pipeline contained in the Builder object. The disadvantage is that (2) isn't as elegant as our current Validator scheme.