Merge pull request #283 from zktuong/master

Fix an issue where serialized AIRR chains were not filtered for `include_fields`.

Author: grst

docs/references.bib

@@ -411,12 +411,12 @@ @article{Lindeman2018
 }
 
 @article{Stephenson2021,
-  doi       = {10.1101/2021.01.13.21249725},
-  url       = {https://doi.org/10.1101/2021.01.13.21249725},
+  doi       = {10.1038/s41591-021-01329-2},
+  url       = {https://www.nature.com/articles/s41591-021-01329-2},
   year      = {2021},
-  month     = jan,
-  publisher = {Cold Spring Harbor Laboratory},
+  month     = apr,
+  publisher = {Springer Science and Business Media {LLC}},
   author    = {Emily Stephenson and Gary Reynolds and Rachel A Botting and Fernando J Calero-Nieto and Michael Morgan and Zewen Kelvin Tuong and Karsten Bach and Waradon Sungnak and Kaylee B Worlock and Masahiro Yoshida and Natsuhiko Kumasaka and Katarzyna Kania and Justin Engelbert and Bayanne Olabi and Jarmila Stremenova Spegarova and Nicola K Wilson and Nicole Mende and Laura Jardine and Louis CS Gardner and Issac Goh and Dave Horsfall and Jim McGrath and Simone Webb and Michael W. Mather and Rik GH Lindeboom and Emma Dann and Ni Huang and Krzysztof Polanski and Elena Prigmore and Florian Gothe and Jonathan Scott and Rebecca P Payne and Kenneth F Baker and Aidan T Hanrath and Ina CD Schim van der Loeff and Andrew S Barr and Amada Sanchez-Gonzalez and Laura Bergamaschi and Federica Mescia and Josephine L Barnes and Eliz Kilich and Angus de Wilton and Anita Saigal and Aarash Saleh and Sam M Janes and Claire M Smith and Nusayhah Gopee and Caroline Wilson and Paul Coupland and Jonathan M Coxhead and Vladimir Y Kiselev and Stijn van Dongen and Jaume Bacardit and Hamish W King and Anthony J Rostron and A John Simpson and Sophie Hambleton and Elisa Laurenti and Paul A Lyons and Kerstin B Meyer and Marko Z Nikolic and Christopher JA Duncan and Ken Smith and Sarah A Teichmann and Menna R Clatworthy and John C Marioni and Berthold Gottgens and Muzlifah Haniffa and},
-  title     = {The cellular immune response to {COVID}-19 deciphered by single cell multi-omics across three {UK} centres},
-  journal   = {BioRxiv}
+  title     = {Single-cell multi-omics analysis of the immune response in {COVID}-19},
+  journal   = {Nature Medicine}
 }

scirpy/io/_datastructures.py

@@ -163,7 +163,9 @@ def add_serialized_chains(self, serialized_chains: str) -> None:
         if not _is_na2(serialized_chains):
             tmp_chains = json.loads(serialized_chains)
             for chain in tmp_chains:
-                self.add_chain(chain)
+                tmp_chain = AirrCell.empty_chain_dict()
+                tmp_chain.update(chain)
+                self.add_chain(tmp_chain)
 
     def _split_chains(self) -> Tuple[bool, dict]:
         """
@@ -234,7 +236,9 @@ def _key_sort_chains(chain) -> Tuple:
         return tuple(-1 if x is None else x for x in sort_tuple)
 
     @staticmethod
-    def _serialize_chains(chains: List[MutableMapping]) -> str:
+    def _serialize_chains(
+        chains: List[MutableMapping], include_fields: Optional[Collection[str]] = None
+    ) -> str:
         """Serialize chains into a JSON object. This is useful for storing
         an arbitrary number of extra chains in a single column of a dataframe."""
         # convert numpy dtypes to python types
@@ -245,7 +249,13 @@ def _serialize_chains(chains: List[MutableMapping]) -> str:
                     chain[k] = chain[k].item()
                 except AttributeError:
                     pass
-        return json.dumps(chains)
+
+        # Filter chains for `include_fields`
+        chains_filtered = [
+            {k: v for k, v in chain.items() if k in include_fields} for chain in chains
+        ]
+
+        return json.dumps(chains_filtered)
 
     def to_airr_records(self) -> Iterable[dict]:
         """Iterate over chains as AIRR-Rearrangent compliant dictonaries.
@@ -292,7 +302,9 @@ def to_scirpy_record(
         include_fields.add("cell_id")
 
         res_dict["multi_chain"], chain_dict = self._split_chains()
-        res_dict["extra_chains"] = self._serialize_chains(chain_dict.pop("extra"))
+        res_dict["extra_chains"] = self._serialize_chains(
+            chain_dict.pop("extra"), include_fields=include_fields
+        )
 
         # add cell-level attributes
         for key in self:

scirpy/io/_io.py

@@ -663,10 +663,12 @@ def to_dandelion(adata: AnnData):
 
 
 @_doc_params(doc_working_model=doc_working_model)
-def from_dandelion(dandelion, transfer=False) -> AnnData:
+def from_dandelion(dandelion, transfer: bool = False, **kwargs) -> AnnData:
     """\
     Import data from `Dandelion <https://github.com/zktuong/dandelion>`_ (:cite:`Stephenson2021`).
 
+    Internally calls :func:`scirpy.io.read_airr`. 
+
     {doc_working_model}
 
     Parameters
@@ -676,6 +678,8 @@ def from_dandelion(dandelion, transfer=False) -> AnnData:
     transfer
         Whether to execute `dandelion.tl.transfer` to transfer all data
         to the :class:`anndata.AnnData` instance.
+    **kwargs
+        Additional arguments passed to :func:`scirpy.io.read_airr`. 
 
     Returns
     -------
@@ -691,7 +695,8 @@ def from_dandelion(dandelion, transfer=False) -> AnnData:
     for col in dandelion_df.columns:
         dandelion_df.loc[dandelion_df[col] == "unassigned", col] = None
 
-    adata = read_airr(dandelion_df)
+    adata = read_airr(dandelion_df, **kwargs)
+
     if transfer:
         ddl.tl.transfer(
             adata, dandelion