Closed pcm32 closed 4 years ago
Thanks! Fixed on the develop branch. I knew about this, but didn't apply it to the symbol case.
Note, this should not happen on real datasets. If you define genes to be symbols, they can not be numbers. If they are numbers, then they are entrez gene IDs. But I've changed it anyways, just to avoid the crash.
Apparently I have run into a case where a gene symbol or id is being intepreted as a numpy int, so I guess we need to be more defensive on that concatenation line:
I have seen this in 0.7.9 as well.
Thanks!