Hi, scanpy has been very helpful in my group’s analyses and has been very efficient.
We are having an issue recently, by concatenating over fifteen studies through scanpy/anndata and we filtered for highly variable genes for all datasets. But we were thinking of an approach that would do the union the top 3,000 highly variable genes in each of the datasets by itself in order to decrease bias. When we did the second approach, the output of the concatenation by the inner parameter returned only 3 genes and the outer returned only 21,000 genes, fewer genes than expected, in a matrix with NaN expression for all cells.
Thanks in advance