Merge pull request #115 from theislab/development

Development

Author: AnnaChristina

.cookietemple.yml

@@ -15,5 +15,5 @@ full_name: Anna Schaar
 email: [email protected]
 project_name: ncem
 project_short_description: ncem.  Learning cell communication from spatial graphs of cells.
-version: 0.1.3
+version: 0.1.4
 license: BSD-3-Clause

.github/release-drafter.yml

@@ -1,5 +1,5 @@
-name-template: "0.1.3 🌈" # <<COOKIETEMPLE_FORCE_BUMP>>
-tag-template: 0.1.3 # <<COOKIETEMPLE_FORCE_BUMP>>
+name-template: "0.1.4 🌈" # <<COOKIETEMPLE_FORCE_BUMP>>
+tag-template: 0.1.4 # <<COOKIETEMPLE_FORCE_BUMP>>
 categories:
     - title: "🚀 Features"
       labels:

.github/workflows/run_tests.yml

@@ -152,7 +152,7 @@ jobs:
                   pip install rich
 
             - name: Download coverage data
-              uses: actions/download-artifact@v2.0.10
+              uses: actions/download-artifact@v3.0.0
               with:
                   name: coverage-data
 

cookietemple.cfg

@@ -1,5 +1,5 @@
 [bumpversion]
-current_version = 0.1.3
+current_version = 0.1.4
 
 [bumpversion_files_whitelisted]
 init_file = ncem/__init__.py

docs/conf.py

@@ -48,10 +48,10 @@
 intersphinx_mapping = dict(
     anndata=("https://anndata.readthedocs.io/en/latest/", None),
     scanpy=("https://scanpy.readthedocs.io/en/latest/", None),
-    numpy=("https://docs.scipy.org/doc/numpy/", None),
-    pandas=("http://pandas.pydata.org/pandas-docs/stable/", None),
-    python=("https://docs.python.org/3", None),
-    scipy=("https://docs.scipy.org/doc/scipy/reference/", None),
+    #numpy=("https://docs.scipy.org/doc/numpy/", None),
+    #pandas=("http://pandas.pydata.org/pandas-docs/stable/", None),
+    #python=("https://docs.python.org/3", None),
+    #scipy=("https://docs.scipy.org/doc/scipy/reference/", None),
 )
 
 # Add any paths that contain templates here, relative to this directory.
@@ -73,9 +73,9 @@
 # the built documents.
 #
 # The short X.Y version.
-version = "0.1.3"
+version = "0.1.4"
 # The full version, including alpha/beta/rc tags.
-release = "0.1.3"
+release = "0.1.4"
 
 # The language for content autogenerated by Sphinx. Refer to documentation
 # for a list of supported languages.

docs/requirements.txt

@@ -1,6 +1,7 @@
+ncem>=0.1.2
 Sphinx>=4.0.1
 sphinx_rtd_theme>=0.5.2
 sphinx-rtd-dark-mode>=1.2.1
 sphinx-automodapi>=0.13
 sphinx_click>=3.0.0
-click>=8.0.1
+click>=7.1.2

ncem/__init__.py

@@ -8,4 +8,4 @@
 __maintainer__ = ", ".join(["Anna C. Schaar", "David S. Fischer"])
 __author__ = ", ".join(["Anna C. Schaar", "David S. Fischer"])
 __email__ = ", ".join(["[email protected]", "[email protected]"])
-__version__ = "0.1.3"
+__version__ = "0.1.4"

ncem/api/train/__init__.py

@@ -1,4 +1,5 @@
 """Initializes a train object in api."""
+import numpy as np
 from ncem.estimators import (Estimator, EstimatorCVAE, EstimatorCVAEncem,
                              EstimatorED, EstimatorEDncem, EstimatorEdNcemNeighborhood, EstimatorGraph,
                              EstimatorInteractions, EstimatorLinear,

ncem/data.py

@@ -21,7 +21,7 @@
 from tqdm import tqdm
 
 
-def get_data_custom(interpreter, n_eval_nodes_per_graph: int=10):
+def get_data_custom(interpreter, deconvolution: bool = False, n_eval_nodes_per_graph: int=10):
     interpreter.undefined_node_types = None
     interpreter.img_to_patient_dict = interpreter.data.celldata.uns["img_to_patient_dict"]
     interpreter.complete_img_keys = list(interpreter.data.img_celldata.keys())
@@ -64,8 +64,10 @@ def get_data_custom(interpreter, n_eval_nodes_per_graph: int=10):
     interpreter.split_data_node(0.1, 0.1)
     interpreter.n_eval_nodes_per_graph = n_eval_nodes_per_graph
     interpreter.cell_names = list(interpreter.data.celldata.uns['node_type_names'].values())
+    if deconvolution:
+        interpreter.proportions = {k: adata.obsm["proportions"] for k, adata in interpreter.data.img_celldata.items()}
+    
 
-
 class GraphTools:
     """GraphTools class."""
 
@@ -1921,6 +1923,109 @@ def _register_graph_features(self, label_selection):
         }
         self.celldata.uns["graph_covariates"] = graph_covariates
 
+        
+class customLoaderDeconvolution(DataLoader):
+    
+    def __init__(
+        self,
+        adata, 
+        patient, 
+        library_id,
+        radius,
+        coord_type='generic',
+        n_rings=1,
+        n_top_genes=None,
+        label_selection=None
+    ):
+        self.adata = adata.copy()
+        self.patient = patient
+        self.library_id = library_id
+
+        print("Loading data from raw files")
+        self.register_celldata(n_top_genes=n_top_genes)
+        self.register_img_celldata()
+        self.register_graph_features(label_selection=label_selection)
+        self.compute_adjacency_matrices(radius=radius, coord_type=coord_type, n_rings=n_rings)
+        self.radius = radius
+
+        print(
+            "Loaded %i images with complete data from %i patients "
+            "over %i cells with %i cell features and %i distinct celltypes."
+            % (
+                len(self.img_celldata),
+                len(self.patients),
+                self.celldata.shape[0],
+                self.celldata.shape[1],
+                len(self.celldata.uns["node_type_names"]),
+            )
+        )
+    
+    def _register_celldata(self, n_top_genes):
+        
+        metadata = {
+            #"cluster_col_preprocessed": self.cluster,
+            "image_col": self.library_id
+        }
+        
+        celldata = self.adata.copy()
+        celldata.uns["metadata"] = metadata
+
+        if self.patient:
+            img_to_patient_dict = {}
+            for p in np.unique(celldata.obs[self.patient]):
+                for i in np.unique(celldata.obs[celldata.obs[self.patient] == p][self.library_id]):
+                    img_to_patient_dict[i] = p
+        else:
+            if self.library_id:
+                img_to_patient_dict = {}
+                for img in np.unique(celldata.obs[self.library_id]):
+                    img_to_patient_dict[str(img)] = "patient"
+            else:
+                img_to_patient_dict = {"image": "patient"}
+        celldata.uns["img_to_patient_dict"] = img_to_patient_dict
+        self.img_to_patient_dict = img_to_patient_dict
+
+        self.celldata = celldata
+        
+    def _register_img_celldata(self):
+        """Load dictionary of of image-wise celldata objects with {imgage key : anndata object of image}."""
+        img_celldata = {}
+        if self.library_id:
+            for k in np.unique(self.celldata.obs[self.library_id]):
+                img_celldata[str(k)] = self.celldata[self.celldata.obs[self.library_id] == k].copy()
+            self.img_celldata = img_celldata
+        else:
+            self.img_celldata = {"image": self.celldata}
+
+    def _register_graph_features(self, label_selection):
+        """Load graph level covariates.
+
+        Parameters
+        ----------
+        label_selection
+            Label selection.
+        """
+        # Save processed data to attributes.
+        for adata in self.img_celldata.values():
+            graph_covariates = {
+                "label_names": {},
+                "label_tensors": {},
+                "label_selection": [],
+                "continuous_mean": {},
+                "continuous_std": {},
+                "label_data_types": {},
+            }
+            adata.uns["graph_covariates"] = graph_covariates
+
+        graph_covariates = {
+            "label_names": {},
+            "label_selection": [],
+            "continuous_mean": {},
+            "continuous_std": {},
+            "label_data_types": {},
+        }
+        self.celldata.uns["graph_covariates"] = graph_covariates
+
 
 class DataLoaderZhang(DataLoader):
     """DataLoaderZhang class. Inherits all functions from DataLoader."""

ncem/interpretation/interpreter.py

@@ -1938,7 +1938,7 @@ def get_sender_receiver_effects(
         interaction_shape = np.int(self.n_features_0**2)
         interactions = np.asarray(dmatrix("target:proportions-1", data))
 
-        y = self.data.celldata.X[self.nodes_idx_all['1'],:]
+        y = self.data.celldata.X
 
         print('using ols parameters.')
         if params_type == 'ols':