which assume that the default is GRCh38. This is true, for now, but when it changes it will be very hard to maintain. I think that such default URLs should be centralized in the respective db module. A general rule should be that no hard coded URLs should be found anywhere but the db package.
Also, testing of the functionality on different builds is not thorough at this point (it is fairly inexistent actually). This should also be fixed...
As the codebase slowly grows, it becomes more important to make sure we have a standard and robust way to handle different genome builds.
I am currently worried about lines like this one:
which assume that the default is
GRCh38
. This is true, for now, but when it changes it will be very hard to maintain. I think that such default URLs should be centralized in the respectivedb
module. A general rule should be that no hard coded URLs should be found anywhere but thedb
package.Also, testing of the functionality on different builds is not thorough at this point (it is fairly inexistent actually). This should also be fixed...