core library (#711)

* move app creation to function

* create engine without load

* flake 8 fixes

* cleanup original scanpy test

* add default config

* handle missing data

* test data changes

* unify update

* load data isn't static anymore

* make app a class

server/app/app.py

@@ -5,27 +5,29 @@ from flask_caching import Cache
 from flask_compress import Compress
 from flask_cors import CORS
 
-from .rest_api.rest import get_api_resources
-from .util.utils import Float32JSONEncoder
-from .web import webapp
+from server.app.rest_api.rest import get_api_resources
+from server.app.util.utils import Float32JSONEncoder
+from server.app.web import webapp
 
-REACTIVE_LIMIT = 1_000_000
 
-app = Flask(__name__, static_folder="web/static")
-app.json_encoder = Float32JSONEncoder
-cache = Cache(app, config={"CACHE_TYPE": "simple", "CACHE_DEFAULT_TIMEOUT": 860_000})
-Compress(app)
-CORS(app)
+class Server:
+    def __init__(self):
+        self.data = None
+        self.cache = Cache(config={"CACHE_TYPE": "simple", "CACHE_DEFAULT_TIMEOUT": 860_000})
 
-# Config
-SECRET_KEY = os.environ.get("CXG_SECRET_KEY", default="SparkleAndShine")
+    def create_app(self):
+        app = Flask(__name__, static_folder="web/static")
+        app.json_encoder = Float32JSONEncoder
+        self.cache.init_app(app)
+        Compress(app)
+        CORS(app)
 
-app.config.update(SECRET_KEY=SECRET_KEY)
+        # Config
+        SECRET_KEY = os.environ.get("CXG_SECRET_KEY", default="SparkleAndShine")
+        app.config.update(SECRET_KEY=SECRET_KEY)
 
-# Application Data
-data = None
-
-resources = get_api_resources()
-app.register_blueprint(webapp.bp)
-app.register_blueprint(resources.blueprint)
-app.add_url_rule("/", endpoint="index")
+        resources = get_api_resources()
+        app.register_blueprint(webapp.bp)
+        app.register_blueprint(resources.blueprint)
+        app.add_url_rule("/", endpoint="index")
+        return app

server/app/driver/driver.py

@@ -11,13 +11,27 @@ Sort order for methods
 
 
 class CXGDriver(metaclass=ABCMeta):
-    def __init__(self, data, args):
-        self.data = self._load_data(data)
-        self.layout_method = args["layout"]
-        self.diffexp_method = args["diffexp"]
-        self.max_category_items = args["max_category_items"]
-        self.diffexp_lfc_cutoff = args["diffexp_lfc_cutoff"]
-        self.cluster = None
+    def __init__(self, data=None, args={}):
+        self.config = self._get_default_config()
+        self.config.update(args)
+        if data:
+            self._load_data(data)
+        else:
+            self.data = None
+
+    def update(self, data=None, args={}):
+        self.config.update(args)
+        if data:
+            self._load_data(data)
+
+    @staticmethod
+    def _get_default_config():
+        return {
+            "layout": None,
+            "diffexp": None,
+            "max_category_items": None,
+            "diffexp_lfc_cutoff": None
+        }
 
     @property
     def features(self):
@@ -27,18 +41,15 @@ class CXGDriver(metaclass=ABCMeta):
             "diffexp": {"available": False},
         }
         # TODO - Interactive limit should be generated from the actual available methods see GH issue #94
-        if self.layout_method:
+        if self.config["layout"]:
             # TODO handle "var" when gene layout becomes available
             features["layout"]["obs"] = {"available": True, "interactiveLimit": 50000}
-        if self.diffexp_method:
+        if self.config["diffexp"]:
             features["diffexp"] = {"available": True, "interactiveLimit": 50000}
-        if self.cluster:
-            features["cluster"] = {"available": True, "interactiveLimit": 50000}
         return features
 
-    @staticmethod
     @abstractmethod
-    def _load_data(data):
+    def _load_data(self, data):
         pass
 
     @abstractmethod

server/app/rest_api/rest.py

@@ -63,7 +63,7 @@ class ConfigAPI(Resource):
                     "dataset": current_app.config["DATASET_TITLE"],
                 },
                 "parameters": {
-                    "max_category_items": current_app.data.max_category_items
+                    "max_category_items": current_app.data.config["max_category_items"]
                 },
                 "library_versions": {
                     "scanpy": pkg_resources.get_distribution("scanpy").version,

server/app/scanpy_engine/scanpy_engine.py

@@ -12,7 +12,7 @@ from server.app.util.errors import (
     PrepareError,
     ScanpyFileError,
 )
-from server.app.util.utils import jsonify_scanpy
+from server.app.util.utils import jsonify_scanpy, requires_data
 from server.app.scanpy_engine.diffexp import diffexp_ttest
 from server.app.util.fbs.matrix import encode_matrix_fbs
 
@@ -27,17 +27,26 @@ Sort order for methods
 
 
 class ScanpyEngine(CXGDriver):
-    def __init__(self, data, args):
+    def __init__(self, data=None, args={}):
         super().__init__(data, args)
-        self._alias_annotation_names(Axis.OBS, args["obs_names"])
-        self._alias_annotation_names(Axis.VAR, args["var_names"])
-        self._validate_data_types()
-        self._validate_data_calculations()
-        self.cell_count = self.data.shape[0]
-        self.gene_count = self.data.shape[1]
-        self.layout_options = ["umap", "tsne"]
-        self.diffexp_options = ["ttest"]
-        self._create_schema()
+        if self.data:
+            self._validate_and_initialize()
+
+    def update(self, data=None, args={}):
+        super().__init__(data, args)
+        if self.data:
+            self._validate_and_initialize()
+
+    @staticmethod
+    def _get_default_config():
+        return {
+            "layout": "umap",
+            "diffexp": "ttest",
+            "max_category_items": 100,
+            "obs_names": None,
+            "var_names": None,
+            "diffexp_lfc_cutoff": 0.01,
+        }
 
     def _alias_annotation_names(self, axis, name):
         """
@@ -95,6 +104,7 @@ class ScanpyEngine(CXGDriver):
                 return True
         return False
 
+    @requires_data
     def _create_schema(self):
         self.schema = {
             "dataframe": {
@@ -128,13 +138,12 @@ class ScanpyEngine(CXGDriver):
                     )
                 self.schema["annotations"][ax].append(ann_schema)
 
-    @staticmethod
-    def _load_data(data):
+    def _load_data(self, data):
         # Based on benchmarking, cache=True has no impact on perf.
         # Note: as of current scanpy/anndata release, setting backed='r' will
         # result in an error.  https://github.com/theislab/anndata/issues/79
         try:
-            result = sc.read(data, cache=True)
+            self.data = sc.read(data, cache=True)
         except ValueError:
             raise ScanpyFileError(
                 "File must be in the .h5ad format. Please read "
@@ -151,8 +160,18 @@ class ScanpyEngine(CXGDriver):
                 f"Error while loading file: {e}, File must be in the .h5ad format, please check "
                 f"that your input and try again."
             )
-        return result
 
+    @requires_data
+    def _validate_and_initialize(self):
+        self._alias_annotation_names(Axis.OBS, self.config["obs_names"])
+        self._alias_annotation_names(Axis.VAR, self.config["var_names"])
+        self._validate_data_types()
+        self._validate_data_calculations()
+        self.cell_count = self.data.shape[0]
+        self.gene_count = self.data.shape[1]
+        self._create_schema()
+
+    @requires_data
     def _validate_data_types(self):
         if self.data.X.dtype != "float32":
             warnings.warn(
@@ -176,7 +195,7 @@ class ScanpyEngine(CXGDriver):
                     )
                 if isinstance(datatype, CategoricalDtype):
                     category_num = len(curr_axis[ann].dtype.categories)
-                    if category_num > 500 and category_num > self.max_category_items:
+                    if category_num > 500 and category_num > self.config['max_category_items']:
                         warnings.warn(
                             f"{str(ax).title()} annotation '{ann}' has {category_num} categories, this may be "
                             f"cumbersome or slow to display. We recommend setting the "
@@ -184,16 +203,17 @@ class ScanpyEngine(CXGDriver):
                             f"annotations with more than 500 categories in the UI"
                         )
 
+    @requires_data
     def _validate_data_calculations(self):
-        layout_key = f"X_{self.layout_method}"
+        layout_key = f"X_{self.config['layout']}"
         try:
             assert layout_key in self.data.obsm_keys()
         except AssertionError:
             raise PrepareError(
-                f"Cannot find a field with coordinates for the {self.layout_method} layout requested. A different"
+                f"Cannot find a field with coordinates for the {self.config['layout']} layout requested. A different"
                 f" layout may have been computed. The requested layout must be pre-calculated and saved "
                 f"back in the h5ad file. You can run "
-                f"`cellxgene prepare --layout {self.layout_method} <datafile>` "
+                f"`cellxgene prepare --layout {self.config['layout']} <datafile>` "
                 f"to solve this problem. "
             )
 
@@ -220,7 +240,7 @@ class ScanpyEngine(CXGDriver):
         mask = np.zeros((count,), dtype=bool)
         for i in filter:
             if type(i) == list:
-                mask[i[0] : i[1]] = True
+                mask[i[0]: i[1]] = True
             else:
                 mask[i] = True
         return mask
@@ -241,6 +261,7 @@ class ScanpyEngine(CXGDriver):
             )
         return mask
 
+    @requires_data
     def _filter_to_mask(self, filter, use_slices=True):
         if use_slices:
             obs_selector = slice(0, self.data.n_obs)
@@ -260,6 +281,7 @@ class ScanpyEngine(CXGDriver):
                 )
         return obs_selector, var_selector
 
+    @requires_data
     def annotation_to_fbs_matrix(self, axis, fields=None):
         if axis == Axis.OBS:
             df = self.data.obs
@@ -269,6 +291,7 @@ class ScanpyEngine(CXGDriver):
             df = df[fields]
         return encode_matrix_fbs(df, col_idx=df.columns)
 
+    @requires_data
     def data_frame_to_fbs_matrix(self, filter, axis):
         """
         Retrieves data 'X' and returns in a flatbuffer Matrix.
@@ -295,6 +318,7 @@ class ScanpyEngine(CXGDriver):
             X = X[:, var_selector]
         return encode_matrix_fbs(X, col_idx=np.nonzero(var_selector)[0], row_idx=None)
 
+    @requires_data
     def diffexp_topN(self, obsFilterA, obsFilterB, top_n=None, interactive_limit=None):
         if Axis.VAR in obsFilterA or Axis.VAR in obsFilterB:
             raise FilterError("Observation filters may not contain vaiable conditions")
@@ -310,7 +334,7 @@ class ScanpyEngine(CXGDriver):
         if top_n is None:
             top_n = DEFAULT_TOP_N
         result = diffexp_ttest(
-            self.data, obs_mask_A, obs_mask_B, top_n, self.diffexp_lfc_cutoff
+            self.data, obs_mask_A, obs_mask_B, top_n, self.config['diffexp_lfc_cutoff']
         )
         try:
             return jsonify_scanpy(result)
@@ -319,6 +343,7 @@ class ScanpyEngine(CXGDriver):
                 "Error encoding differential expression to JSON"
             )
 
+    @requires_data
     def layout_to_fbs_matrix(self):
         """
         Return the default 2-D layout for cells as a FBS Matrix.
@@ -328,14 +353,14 @@ class ScanpyEngine(CXGDriver):
         * only returns Matrix in columnar layout
         """
         try:
-            full_embedding = self.data.obsm[f"X_{self.layout_method}"]
+            full_embedding = self.data.obsm[f"X_{self.config['layout']}"]
             if full_embedding.shape[1] > 2:
                 warnings.warn(f"Warning: found {full_embedding.shape[1]} \
                 components of embedding. Using the first two for layout display.")
             df_layout = full_embedding[:, :2]
         except ValueError as e:
             raise PrepareError(
-                f"Layout has not been calculated using {self.layout_method}, "
+                f"Layout has not been calculated using {self.config['layout']}, "
                 f"please prepare your datafile and relaunch cellxgene") from e
 
         normalized_layout = (df_layout - df_layout.min()) / (df_layout.max() - df_layout.min())

server/app/util/constants.py

@@ -30,3 +30,4 @@ class DiffExpMode(AugmentedEnum):
 JSON_NaN_to_num_warning_msg = (
     "JSON encoding failure - please verify all data are finite values (no NaN or Infinities)"
 )
+REACTIVE_LIMIT = 1_000_000

server/app/util/errors.py

@@ -50,3 +50,12 @@ class ScanpyFileError(Exception):
 
     def __init__(self, message):
         self.message = message
+
+
+class DriverError(Exception):
+    """
+    Raised when file loaded into scanpy is misformatted
+    """
+
+    def __init__(self, message):
+        self.message = message

server/app/util/utils.py

@@ -1,6 +1,10 @@
+from functools import wraps
+
 from flask import json
 from numpy import float32, integer
 
+from server.app.util.errors import DriverError
+
 
 class Float32JSONEncoder(json.JSONEncoder):
     def __init__(self, *args, **kwargs):
@@ -28,3 +32,12 @@ def custom_format_warning(msg, *args, **kwargs):
 
 def jsonify_scanpy(data):
     return json.dumps(data, cls=Float32JSONEncoder, allow_nan=False)
+
+
+def requires_data(func):
+    @wraps(func)
+    def wrapped_function(self, *args, **kwargs):
+        if self.data is None:
+            raise DriverError(f"error data must be loaded before you call {func.__name__}")
+        return func(self, *args, **kwargs)
+    return wrapped_function

server/cli/launch.py

@@ -7,6 +7,7 @@ import webbrowser
 
 import click
 
+from server.app.app import Server
 from server.app.util.errors import ScanpyFileError
 from server.app.util.utils import custom_format_warning
 
@@ -116,8 +117,9 @@ def launch(
     cellxgene_url = f"http://{host}:{port}"
 
     # Import Flask app
-    from server.app.app import app
+    server = Server()
 
+    app = server.create_app()
     app.config.update(DATASET_TITLE=title)
 
     if not verbose:

server/test/test_scanpy_engine.py

@@ -12,7 +12,7 @@ from server.app.scanpy_engine.scanpy_engine import ScanpyEngine
 from server.app.util.errors import FilterError
 
 
-class UtilTest(unittest.TestCase):
+class EngineTest(unittest.TestCase):
     def setUp(self):
         args = {
             "layout": "umap",
@@ -22,9 +22,7 @@ class UtilTest(unittest.TestCase):
             "var_names": None,
             "diffexp_lfc_cutoff": 0.01,
         }
-
         self.data = ScanpyEngine("example-dataset/pbmc3k.h5ad", args)
-        self.data._create_schema()
 
     def test_init(self):
         self.assertEqual(self.data.cell_count, 2638)

server/test/test_scanpy_engine_data_load.py

@@ -0,0 +1,50 @@
+import unittest
+import json
+
+from server.app.scanpy_engine.scanpy_engine import ScanpyEngine
+from server.app.util.errors import DriverError
+
+
+class DataLoadEngineTest(unittest.TestCase):
+    def setUp(self):
+        self.data_file = "example-dataset/pbmc3k.h5ad"
+        self.data = ScanpyEngine()
+
+    def test_init(self):
+        self.assertIsNone(self.data.data)
+
+    def test_delayed_load_args(self):
+        args = {
+            "layout": "tsne",
+            "diffexp": "ttest",
+            "max_category_items": 1000,
+            "obs_names": "foo",
+            "var_names": "bar",
+            "diffexp_lfc_cutoff": 0.1,
+        }
+        self.data.update(args=args)
+        self.assertEqual(args, self.data.config)
+
+    def test_requires_data(self):
+        with self.assertRaises(DriverError):
+            self.data._create_schema()
+
+    def test_delayed_load_data(self):
+        self.data.update(data=self.data_file)
+        self.data._create_schema()
+        self.assertEqual(self.data.cell_count, 2638)
+        self.assertEqual(self.data.gene_count, 1838)
+        epsilon = 0.000_005
+        self.assertTrue(self.data.data.X[0, 0] - -0.171_469_51 < epsilon)
+
+    def test_diffexp_topN(self):
+        self.data.update(data=self.data_file)
+        f1 = {"filter": {"obs": {"index": [[0, 500]]}}}
+        f2 = {"filter": {"obs": {"index": [[500, 1000]]}}}
+        result = json.loads(self.data.diffexp_topN(f1["filter"], f2["filter"]))
+        self.assertEqual(len(result), 10)
+        result = json.loads(self.data.diffexp_topN(f1["filter"], f2["filter"], 20))
+        self.assertEqual(len(result), 20)
+
+    if __name__ == "__main__":
+        unittest.main()