ImplicitModificationWarning: Trying to modify attribute `._uns` of view, initializing view as actual. self.data[key] = value

• I have checked that this issue has not already been reported.
• I have confirmed this bug exists on the latest version of scanpy.
• (optional) I have confirmed this bug exists on the master branch of scanpy.

#remove all ribosomal genes
malat1 = adata1.var_names.str.startswith('MALAT1')
ribo_genes = adata1.var_names.str.startswith(("RPS","RPL"))
# we need to redefine the mito_genes since they were first 
# calculated on the full object before removing low expressed genes.
mito_genes = adata1.var_names.str.startswith('MT-')
hb_genes = adata1.var_names.str.contains('^HB[^(P)]')

remove = np.add(mito_genes, malat1)
remove = np.add(remove, hb_genes)
keep = np.invert(remove)

adata3 = adata1[:,keep]

adata3

Versions

anndata 0.8.0
scanpy 1.7.2

I am getting this error "ImplicitModificationWarning: Trying to modify attribute ._uns of view, initializing view as actual.
self.data[key] = value " as soon as i write the above code(subset data based on genes i need) and then the kernel crashes because of memory issues

Start by reproducing the reported gene-subsetting snippet with the listed Scanpy and anndata versions, then trace the view initialization and memory behavior in the relevant code paths. Done means the operation no longer produces the reported warning followed by a kernel crash, with the behavior verified against an appropriate regression test.