Merge pull request #74 from klarman-cell-observatory/boli

Added obsp and varp

pegasusio/citeseq_data.py

@@ -25,10 +25,12 @@ def __init__(
         metadata: dict,
         barcode_multiarrays: Dict[str, np.ndarray] = None,
         feature_multiarrays: Dict[str, np.ndarray] = None,
+        barcode_multigraphs: Dict[str, csr_matrix] = None,
+        feature_multigraphs: Dict[str, csr_matrix] = None,
         cur_matrix: str = "raw.count",
     ) -> None:
         assert metadata["modality"] == "citeseq"
-        super().__init__(barcode_metadata, feature_metadata, matrices, metadata, barcode_multiarrays, feature_multiarrays, cur_matrix)
+        super().__init__(barcode_metadata, feature_metadata, matrices, metadata, barcode_multiarrays, feature_multiarrays, barcode_multigraphs, feature_multigraphs, cur_matrix)
 
 
     def from_anndata(self, data: anndata.AnnData, genome: str = None, modality: str = None) -> None:

pegasusio/cyto_data.py

@@ -1,5 +1,6 @@
 import numpy as np
 import pandas as pd
+from scipy.sparse import csr_matrix
 from typing import List, Dict, Union
 
 import logging
@@ -23,10 +24,12 @@ def __init__(
         metadata: dict,
         barcode_multiarrays: Dict[str, np.ndarray] = None,
         feature_multiarrays: Dict[str, np.ndarray] = None,
+        barcode_multigraphs: Dict[str, csr_matrix] = None,
+        feature_multigraphs: Dict[str, csr_matrix] = None,
         cur_matrix: str = "raw.data",
     ) -> None:
         assert metadata["modality"] == "cyto"
-        super().__init__(barcode_metadata, feature_metadata, matrices, metadata, barcode_multiarrays, feature_multiarrays, cur_matrix)
+        super().__init__(barcode_metadata, feature_metadata, matrices, metadata, barcode_multiarrays, feature_multiarrays, barcode_multigraphs, feature_multigraphs, cur_matrix)
 
 
     def from_anndata(self, data: anndata.AnnData, genome: str = None, modality: str = None) -> None:

pegasusio/hdf5_utils.py

@@ -201,6 +201,13 @@ def load_loom_file(input_loom: str, genome: str = None, modality: str = None) ->
             else:
                 raise ValueError(f"Detected row attribute '{key}' has ndim = {arr.ndim}!")
 
+        barcode_multigraphs = {}
+        for key, graph in ds.col_graphs.items():
+            barcode_multigraphs[key] = csr_matrix(graph)
+        feature_multigraphs = {}
+        for key, graph in ds.row_graphs.items():
+            feature_multigraphs[key] = csr_matrix(graph)
+
         matrices = {}
         for key, mat in ds.layers.items():
             key = "X" if key == "" else key
@@ -220,7 +227,7 @@ def load_loom_file(input_loom: str, genome: str = None, modality: str = None) ->
             else:
                 metadata["modality"] = "rna"
 
-        unidata = UnimodalData(barcode_metadata, feature_metadata, matrices, metadata, barcode_multiarrays, feature_multiarrays)
+        unidata = UnimodalData(barcode_metadata, feature_metadata, matrices, metadata, barcode_multiarrays, feature_multiarrays, barcode_multigraphs, feature_multigraphs)
         unidata.separate_channels()
 
     data = MultimodalData(unidata)
@@ -239,13 +246,20 @@ def write_loom_file(data: MultimodalData, output_file: str) -> None:
     if len(matrices) == 0:
         raise ValueError("Could not write empty matrix to a loom file!")
 
+    def _replace_slash(name: str) -> str:
+        """ Replace slash with |
+        """
+        if name.find('/') >= 0:
+            return name.replace('/', '|')
+        return name
+
     def _process_attrs(key_name: str, attrs: pd.DataFrame, attrs_multi: dict) -> Dict[str, object]:
         res_dict = {key_name: attrs.index.values}
         for key in attrs.columns:
-            res_dict[key] = np.array(attrs[key].values)
+            res_dict[_replace_slash(key)] = np.array(attrs[key].values)
         for key, value in attrs_multi.items():
             if value.ndim > 1: # value.ndim == 1 refers to np.recarray, which will not be written to a loom file.
-                res_dict[key] = value if value.shape[1] > 1 else value[:, 0]
+                res_dict[_replace_slash(key)] = value if value.shape[1] > 1 else value[:, 0]
         return res_dict
 
     row_attrs = _process_attrs("Gene", data.var, data.varm)
@@ -262,9 +276,16 @@ def _process_attrs(key_name: str, attrs: pd.DataFrame, attrs_multi: dict) -> Dic
     file_attrs = {}
     for key, value in data.uns.items():
         if isinstance(value, str):
-            file_attrs[key] = value
+            file_attrs[_replace_slash(key)] = value
 
     import loompy
     loompy.create(output_file, layers, row_attrs, col_attrs, file_attrs = file_attrs)
 
+    if len(data.varp) > 0 or len(data.obsp) > 0:
+        with loompy.connect(output_file) as ds:
+            for key, value in data.varp.items():
+                ds.row_graphs[_replace_slash(key)] = value
+            for key, value in data.obsp.items():
+                ds.col_graphs[_replace_slash(key)] = value
+
     logger.info(f"{output_file} is written.")

pegasusio/multimodal_data.py

@@ -110,6 +110,24 @@ def varm(self, varm: Dict[str, np.ndarray]):
         assert self._unidata is not None
         self._unidata.varm = varm
 
+    @property
+    def obsp(self) -> Union[Dict[str, csr_matrix], None]:
+        return self._unidata.obsp if self._unidata is not None else None
+
+    @obsp.setter
+    def obsp(self, obsp: Dict[str, csr_matrix]):
+        assert self._unidata is not None
+        self._unidata.obsp = obsp
+
+    @property
+    def varp(self) -> Dict[str, csr_matrix]:
+        return self._unidata.varp if self._unidata is not None else None
+
+    @varp.setter
+    def varp(self, varp: Dict[str, csr_matrix]):
+        assert self._unidata is not None
+        self._unidata.varp = varp
+
     @property
     def uns(self) -> Union[dict, None]:
         return self._unidata.uns if self._unidata is not None else None
@@ -128,6 +146,18 @@ def shape(self, _shape: Tuple[int, int]):
         assert self._unidata is not None
         self._unidata.shape = _shape
 
+    def get_attr_type(self, attr:str) -> str:
+        """ Surrogate function to return registered type for an attribute
+        """
+        assert self._unidata is not None
+        self._unidata.get_attr_type(attr)
+
+    def register_attr(self, attr: str, attr_type: str = None) -> None:
+        """ Surrogate function to register an attribute (either in obs or obsm) with an attr_type (e.g. signature, cluster, basis)
+        """
+        assert self._unidata is not None
+        self._unidata.register_attr(attr, attr_type)
+
     def as_float(self, matkey: str = None) -> None:
         """ Surrogate function to convert matrix to float """
         assert self._unidata is not None
@@ -393,7 +423,7 @@ def filter_data(self,
 
 
     def concat_data(self, modality: str = "rna"):
-        """ Used for raw data, Ignore multiarrays and only consider one matrix per unidata """
+        """ Used for raw data, Ignore multiarrays/multigraphs and only consider one matrix per unidata """
         genomes = []
         unidata_arr = []
 

pegasusio/nanostring_data.py

@@ -1,5 +1,6 @@
 import numpy as np
 import pandas as pd
+from scipy.sparse import csr_matrix
 from typing import List, Dict, Union
 
 import logging
@@ -21,10 +22,12 @@ def __init__(
         metadata: dict,
         barcode_multiarrays: Dict[str, np.ndarray] = None,
         feature_multiarrays: Dict[str, np.ndarray] = None,
+        barcode_multigraphs: Dict[str, csr_matrix] = None,
+        feature_multigraphs: Dict[str, csr_matrix] = None,
         cur_matrix: str = "Q3Norm",
     ) -> None:
         assert metadata["modality"] == "nanostring"
-        super().__init__(barcode_metadata, feature_metadata, matrices, metadata, barcode_multiarrays, feature_multiarrays, cur_matrix)
+        super().__init__(barcode_metadata, feature_metadata, matrices, metadata, barcode_multiarrays, feature_multiarrays, barcode_multigraphs, feature_multigraphs, cur_matrix)
 
 
     def from_anndata(self, data: anndata.AnnData, genome: str = None, modality: str = None) -> None: