Hi,
I used to concatenate datasets to a reference one using the bound method a.concat(b). It was fairly handy because var were merged in an intersect fashion while obs were merged in an outer fashion, with NA added when a column was specific to one or the other dataset. This method will be deprecated in a future version.
But when I use anndata.concat([a, b], join=“inner”) instead, my var are indeed the intersect of the vars of the objects, but it is the same for obs. If I’m not mistaken this is not the default behavior of the method that will be deprecated…
For instance, I wish I could keep values of my annotation from the first anndata and put NAs for obs of other datasets when the column does not exist.
I tried every merge option and none gave this result.
Someone has got an idea ?
Best,
C.