Allow passing a h5ad object directly into the constructor

inferelator/benchmarking/scenic.py

@@ -67,6 +67,9 @@ class SCENICWorkflow(SingleCellWorkflow):
 
     do_scenic = True
 
+    _do_preprocessing = True
+    _do_scaling = True
+
     dask_temp_path = None
     _tmp_handle = None
 
@@ -119,7 +122,8 @@ def startup_finish(self):
 
             sc.pp.log1p(self.data._adata)
 
-        sc.pp.scale(self.data._adata, max_value=10)
+        if self._do_scaling:
+            sc.pp.scale(self.data._adata, max_value=10)
 
     def create_feather_file_from_prior(self):
 

inferelator/tests/test_data_loader.py

@@ -3,6 +3,7 @@
 import os
 import tempfile
 import pandas as pd
+import anndata as ad
 import numpy as np
 import numpy.testing as npt
 import pandas.testing as pdt
@@ -35,6 +36,14 @@ def test_h5ad(self):
 
         npt.assert_array_almost_equal(data.values, self.worker.data.expression_data)
 
+    def test_h5ad_obj(self):
+        _, data = test_prebuilt.counts_yeast_single_cell_chr01(filetype='h5ad')
+
+        self.worker.set_expression_file(h5ad=ad.AnnData(data))
+        self.worker.read_expression()
+
+        npt.assert_array_almost_equal(data.values, self.worker.data.expression_data)
+
     def test_hdf5(self):
         file, data = test_prebuilt.counts_yeast_single_cell_chr01(filetype='hdf5')
 

inferelator/utils/loader.py

@@ -77,16 +77,25 @@ def load_data_h5ad(
         :rtype: InferelatorData
         """
 
+        if isinstance(h5ad_file, anndata.AnnData):
+            data = h5ad_file
+            Debug.vprint(
+                f"Using AnnData object {data.shape} "
+                f"(Layer: {use_layer if use_layer is not None else 'X'})",
+                level=0
+            )
+
         # Read H5AD file
-        Debug.vprint(
-            f"Loading AnnData file {h5ad_file} "
-            f"(Layer: {use_layer if use_layer is not None else 'X'})",
-            level=0
-        )
+        else:
+            Debug.vprint(
+                f"Loading AnnData file {h5ad_file} "
+                f"(Layer: {use_layer if use_layer is not None else 'X'})",
+                level=0
+            )
 
-        with warnings.catch_warnings():
-            warnings.simplefilter("ignore", anndata.OldFormatWarning)
-            data = anndata.read_h5ad(self.input_path(h5ad_file))
+            with warnings.catch_warnings():
+                warnings.simplefilter("ignore", anndata.OldFormatWarning)
+                data = anndata.read_h5ad(self.input_path(h5ad_file))
 
         # Read in meta data
         if meta_data_file is None and data.obs.shape[1] > 0:
@@ -113,28 +122,24 @@ def load_data_h5ad(
                 f"Layer {use_layer} is not in {h5ad_file}"
             )
 
-        # Build an InferelatorData object from a layer
-        elif use_layer is not None:
-            data = InferelatorData(
-                data.layers[use_layer].copy() if use_layer != "X" else data.X,
-                gene_names=data.var_names.copy(),
-                sample_names=data.obs_names.copy(),
-                meta_data=pd.concat((data.obs, meta_data), axis=1),
-                gene_data=pd.concat((data.var, gene_metadata), axis=1)
-            )
+        if use_layer != "X":
+            data_ref = data.layers[use_layer]
+        else:
+            data_ref = data.X
 
         # Build an InferelatorData object from everything
-        elif keep_other_data:
+        if keep_other_data:
+            data.X = data_ref
             data = InferelatorData(
                 data,
                 meta_data=meta_data,
                 gene_data=gene_metadata
             )
 
-        # Build an InferelatorData object from only the `X` data
+        # Build an InferelatorData object from a layer
         else:
             data = InferelatorData(
-                data.X.copy(),
+                data_ref.copy(),
                 gene_names=data.var_names.copy(),
                 sample_names=data.obs_names.copy(),
                 meta_data=pd.concat((data.obs, meta_data), axis=1),

inferelator/workflow.py

@@ -1,13 +1,13 @@
 """
-Construct inferelator workflows from preprocessing, postprocessing, and regression modules
+Construct inferelator workflows from preprocessing,
+postprocessing, and regression modules
 """
 
 import inspect
 
 from inferelator.utils import is_string
 from inferelator.workflows.workflow_base import (
     WorkflowBase,
-    WorkflowBaseLoader
 )
 from inferelator.regression.base_regression import _RegressionWorkflowMixin
 
@@ -151,7 +151,6 @@ class InferelatorRegressWorkflow(workflow_class):
 
             regression_class = AMUSRRegressionWorkflowMixin
 
-
         elif regression == "stars":
 
             from inferelator.regression.stability_selection import (

inferelator/workflows/workflow_base.py

@@ -275,7 +275,7 @@ def set_expression_file(
             self._expression_loader = _HDF5
             self._h5_layer = h5_layer
         elif h5ad is not None:
-            self._set_file_name("expression_matrix_file", h5ad)
+            self._set_expression("expression_matrix_file", h5ad)
             self._expression_loader = _H5AD
             self._h5_layer = h5_layer
         elif mtx is not None:
@@ -488,7 +488,7 @@ def _set_file_name(
         """
 
         if file_name is None:
-            return
+            return None
 
         if file_name not in self._file_format_settings:
             self._file_format_settings[file_name] = copy.copy(
@@ -498,6 +498,27 @@ def _set_file_name(
         self._check_file_exists(file_name)
         self._set_with_warning(attr_name, file_name)
 
+    def _set_expression(
+        self,
+        attr_name,
+        file_or_object
+    ):
+
+        if file_or_object is None:
+            return None
+
+        if isinstance(file_or_object, str):
+            self._set_file_name(
+                attr_name,
+                file_or_object
+            )
+
+        else:
+            self._set_with_warning(
+                attr_name,
+                file_or_object
+            )
+
     def _check_file_exists(
         self,
         file_name