Does each mother-cell have either two daughters or no daughter (right-censored or dying)?
Are the last frame a mother is observed and the first frames its daughters are observed consistent?
Is the right-censoring flag consistent with no daughters?
Potentially some further sanity checks for specific formats:
some formats assign track-names, common to all cells related to each other, which should then be double-checked.
Some have explicit entries to indicate divisions, which should be consistent with the birth of the two daughters.
Some can have naming-conventions for related cells, which should be double-checked.
If any of these inconsistencies occurs, by default all cells related to the cells involved should be discarded with a warning.
We should provide options to fix some common cases, if the user explicitly requests to do so (via an input option):
If the two sisters are not first observed in the same frame, the later one can be forward-dated.
If the mother has only one observed daughter, the second can be made up. In this case, it would be a never-observed cell with unknown fate (ie fifth column is '1').
Add sanity checks for any input track files:
Potentially some further sanity checks for specific formats:
If any of these inconsistencies occurs, by default all cells related to the cells involved should be discarded with a warning.
We should provide options to fix some common cases, if the user explicitly requests to do so (via an input option):