The spec defines that the chromosome lengths should be stored as a 32bit integer. This works well for human but could pose problems for plant genomes. On the other hand is given a lot of freedom with the data types:
The column data types are given as numpy equivalents. They are only defaults and may be altered as desired.
While this allows a user to store the chromosome lengths using 64bit instead technically they are also allow to store them as strings. I don't think that was intended. Here is my suggestion for a tighter definition:
The column data types are given as numpy equivalents. The data types may be changed in width as desired.
The spec defines that the chromosome lengths should be stored as a 32bit integer. This works well for human but could pose problems for plant genomes. On the other hand is given a lot of freedom with the data types:
While this allows a user to store the chromosome lengths using 64bit instead technically they are also allow to store them as strings. I don't think that was intended. Here is my suggestion for a tighter definition: