Fix that pl.clonotype_network couldn't use non-standard obsm key (#545)

* Fix that pl.clonotype_network couldn't use non-standard obsm key

* Update changelog

* fix usage of _graph_from_coordinates in all instances

CHANGELOG.md

@@ -10,11 +10,18 @@ and this project adheres to [Semantic Versioning][].
 
 ## [Unreleased]
 
-### Addition
+### Additions
 
 - Isotypically included B cells are now labelled as `receptor_subtype="IGH+IGK/L"` instead of `ambiguous` in `tl.chain_qc`. ([#537](https://github.com/scverse/scirpy/pull/537))
-- Added the `normalized_hamming` metric to `pp.ir_dist` that accounts for differences in CDR3 sequence length. Additionally,
- the hamming distance was reimplemented with numba, achieving a significant speedup ([#512](https://github.com/scverse/scirpy/pull/512)).
+- Added the `normalized_hamming` metric to `pp.ir_dist` that accounts for differences in CDR3 sequence length ([#512](https://github.com/scverse/scirpy/pull/512)).
+
+### Performance improvements
+
+- The hamming distance was reimplemented with numba, achieving a significant speedup ([#512](https://github.com/scverse/scirpy/pull/512)).
+
+### Fixes
+
+- Fix that pl.clonotype_network couldn't use non-standard obsm key ([#545](https://github.com/scverse/scirpy/pull/545)).
 
 ### Other changes
 

src/scirpy/pl/_clonotypes.py

@@ -197,7 +197,7 @@ def clonotype_network(
  pass
 
  clonotype_res = params.adata.uns[clonotype_key]
- coords, adj_mat = _graph_from_coordinates(params.adata, clonotype_key)
+ coords, adj_mat = _graph_from_coordinates(params.adata, clonotype_key, basis)
  nx_graph = nx.Graph(_distance_to_connectivity(adj_mat))
  # in 2.6 networkx added functionality to draw self-loops. We don't want
  # them plotted, so we remove them here

src/scirpy/tests/conftest.py

@@ -196,12 +196,16 @@ def adata_define_clonotype_clusters_singletons():
 
 
 @pytest.fixture
-def adata_clonotype_network(adata_conn):
+def adata_clonotype_network(adata_conn, request):
  """Adata with clonotype network computed.
 
  adata derived from adata_conn that also contains some gene expression data
  for plotting.
  """
+ try:
+ kwargs = request.param
+ except AttributeError:
+ kwargs = {}
  if isinstance(adata_conn, AnnData):
  adata = AnnData(
  var=pd.DataFrame().assign(gene_symbol=["CD8A", "CD4"]).set_index("gene_symbol"),
@@ -216,7 +220,7 @@ def adata_clonotype_network(adata_conn):
  obsm=adata_conn.obsm,
  )
  adata.obs["continuous"] = [3, 4, 0, 0, 7, 14, 1, 0, 2, 2, 0]
- ir.tl.clonotype_network(adata, sequence="aa", metric="alignment")
+ ir.tl.clonotype_network(adata, sequence="aa", metric="alignment", **kwargs)
  return adata
  else:
  adata_gex = AnnData(
@@ -231,7 +235,7 @@ def adata_clonotype_network(adata_conn):
  )
  mdata = MuData({"gex": adata_gex, "airr": adata_conn.mod["airr"]})
  mdata.obs["continuous"] = [3, 4, 0, 0, 7, 14, 1, 0, 2, 2, 0]
- ir.tl.clonotype_network(mdata, sequence="aa", metric="alignment")
+ ir.tl.clonotype_network(mdata, sequence="aa", metric="alignment", **kwargs)
  return mdata
 
 

src/scirpy/tests/test_plotting.py

@@ -115,18 +115,18 @@ def test_clonotype_modularity(adata_clonotype_modularity, jitter, show_size_lege
  )
 
 
+@pytest.mark.parametrize(
+ "adata_clonotype_network,kwargs",
+ [[{}, {}], [{"key_added": "foo"}, {"basis": "foo"}]],
+ indirect=["adata_clonotype_network"],
+)
 @pytest.mark.parametrize("color_by_n_cells", [True, False])
 @pytest.mark.parametrize("scale_by_n_cells", [True, False])
 @pytest.mark.parametrize("show_size_legend", [True, False])
 @pytest.mark.parametrize("show_legend", [True, False])
 @pytest.mark.parametrize("show_labels", [True, False])
 def test_clonotype_network(
- adata_clonotype_network,
- color_by_n_cells,
- scale_by_n_cells,
- show_size_legend,
- show_legend,
- show_labels,
+ adata_clonotype_network, color_by_n_cells, scale_by_n_cells, show_size_legend, show_legend, show_labels, kwargs
 ):
  adata = adata_clonotype_network
  p = pl.clonotype_network(
@@ -136,6 +136,7 @@ def test_clonotype_network(
  show_size_legend=show_size_legend,
  show_legend=show_legend,
  show_labels=show_labels,
+ **kwargs,
  )
  assert isinstance(p, plt.Axes)
 

src/scirpy/tl/_clonotypes.py

@@ -579,7 +579,7 @@ def clonotype_network(
  return coord_df
 
 
-def _graph_from_coordinates(adata: AnnData, clonotype_key: str) -> tuple[pd.DataFrame, sp.csr_matrix]:
+def _graph_from_coordinates(adata: AnnData, clonotype_key: str, basis: str) -> tuple[pd.DataFrame, sp.csr_matrix]:
  """
  Given an AnnData object on which `tl.clonotype_network` was ran, and
  the corresponding `clonotype_key`, extract a data-frame
@@ -600,7 +600,7 @@ def _graph_from_coordinates(adata: AnnData, clonotype_key: str) -> tuple[pd.Data
 
  # Retrieve coordinates and reduce them to one coordinate per node
  coords = (
- cast(pd.DataFrame, adata.obsm["X_clonotype_network"])
+ cast(pd.DataFrame, adata.obsm[f"X_{basis}"])
  .dropna(axis=0, how="any")
  .join(dist_idx_lookup)
  .join(clonotype_label_lookup)
@@ -653,7 +653,7 @@ def clonotype_network_igraph(
  raise KeyError(f"{basis} not found in `adata.uns`. Did you run `tl.clonotype_network`?") from None
  if f"X_{basis}" not in params.adata.obsm_keys():
  raise KeyError(f"X_{basis} not found in `adata.obsm`. Did you run `tl.clonotype_network`?")
- coords, adj_mat = _graph_from_coordinates(params.adata, clonotype_key)
+ coords, adj_mat = _graph_from_coordinates(params.adata, clonotype_key, basis)
 
  graph = igraph_from_sparse_matrix(adj_mat, matrix_type="distance")
  # flip y axis to be consistent with networkx