Merge pull request #41 from KrishnaswamyLab/dev

Dev

Author: dburkhardt

.travis.yml

@@ -1,6 +1,5 @@
 language: python
 python:
-  - '3.5'
   - '3.6'
   - '3.7'
   - '3.8'

comparison/comparison.py

@@ -188,20 +188,33 @@ def fit_kNN(self, data, **kwargs):
         self.graph_knn = gt.Graph(data, n_pca=100,  kernel_symm=None, use_pygsp=True,
                                 random_state=self.seed, **kwargs)
 
-    def generate_ground_truth_pdf(self, data_phate=None):
-        '''Takes a set of PHATE coordinates over a set of points and creates an underlying
-        ground truth pdf over the points as a convex combination of the input phate coords.
-        '''
-        np.random.seed(self.seed)
+    def fit_phate(self, data, **kwargs):
+        try:
+            import phate
+        except ModuleNotFoundError:
+            raise ModuleNotFoundError('PHATE must be installed. Install via pip `pip install --user phate`')
 
-        if data_phate is not None:
-            self.data_phate = data_phate
+        self.set_phate(phate.PHATE(n_components=3, **kwargs).fit_transform(data))
+        return self.data_phate
 
+    def set_phate(self, data_phate):
         if not data_phate.shape[1] == 3:
             raise ValueError('data_phate must have 3 dimensions')
         if not np.isclose(data_phate.mean(), 0):
             # data_phate must be mean-centered
             data_phate = scipy.stats.zscore(data_phate, axis=0)
+        self.data_phate = data_phate
+
+    def generate_ground_truth_pdf(self, data_phate=None):
+        '''Takes a set of PHATE coordinates over a set of points and creates an underlying
+        ground truth pdf over the points as a convex combination of the input phate coords.
+        '''
+        np.random.seed(self.seed)
+
+        if data_phate is not None:
+            self.set_phate(data_phate)
+        elif self.data_phate is None:
+            raise ValueError('data_phate must be set prior to running generate_ground_truth_pdf().')
 
         # Create an array of values that sums to 1
         data_simplex = np.sort(np.random.uniform(size=(2)))

data/Klein_parameter_search_results.pickle.gz

@@ -4,7 +4,7 @@
 # You can set these variables from the command line.
 SPHINXOPTS    =
 SPHINXBUILD   = sphinx-build
-SPHINXPROJ    = PHATE
+SPHINXPROJ    = MELD
 SOURCEDIR     = source
 BUILDDIR      = build
 
@@ -17,4 +17,4 @@ help:
 # Catch-all target: route all unknown targets to Sphinx using the new
 # "make mode" option.  $(O) is meant as a shortcut for $(SPHINXOPTS).
 %: Makefile
-	@$(SPHINXBUILD) -M $@ "$(SOURCEDIR)" "$(BUILDDIR)" $(SPHINXOPTS) $(O)
+	@$(SPHINXBUILD) -M $@ "$(SOURCEDIR)" "$(BUILDDIR)" $(SPHINXOPTS) $(O)

doc/Makefile

@@ -3,6 +3,6 @@ scipy>=1.1.0
 pandas>=0.25
 future
 graphtools>=0.1.8.1
-sphinx<=1.8.5
+sphinx
 sphinxcontrib-napoleon
 autodocsumm

doc/source/requirements.txt

@@ -3,3 +3,4 @@
 from .meld import MELD
 from .cluster import VertexFrequencyCluster
 from .version import __version__
+from .benchmark import Benchmarker

meld/__init__.py

@@ -0,0 +1,198 @@
+# Copyright (C) 2020 Krishnaswamy Lab, Yale University
+
+import numpy as np
+import scipy
+import sklearn
+import meld
+import graphtools as gt
+
+
+class Benchmarker(object):
+    """Creates random signals over a dataset for benchmarking.
+
+    Results are used for quantitative comparisons and for parameter searches
+    on a specific dataset.
+
+    Parameters
+    ----------
+    seed : integer or numpy.RandomState, optional, default: None
+        Random state. Defaults to the global `numpy` random number generator
+
+    Attributes
+    ----------
+    data_phate : array, shape=[n_samples, 3]
+        Embedding of the data used to create random signals
+    pdf : array, shape=[n_samples]
+        Ground truth probability density function created over the input data.
+    RES_int : array, shape=[n_samples]
+        An integer representation of the RES used for k-NN and graph averaging.
+    RES : array, shape=[n_samples, 2]
+        Raw Experimental Signal (RES) as described in Burkhardt et al. (2020).
+        Indicates the sample to which each cell is assigned.
+    graph : graphtools.base.BaseGraph
+        The graph built on the input data
+    graph_kNN : graphtools.graphs.kNNGraph
+        The graph built on the input data
+    meld_op : meld.meld.MELD
+        MELD operator used to derive an EES
+    EES : array, shape=[n_samples, 2]
+        Enhanced Experimental Signal (EES) a conditional probability that a cell was
+        originally sampled from either condition. Should converge to Benchmarker.pdf
+
+    """
+    def __init__(self, seed=None):
+        self.seed = seed
+        self.data_phate = None
+        self.pdf = None
+        self.RES_int = None
+        self.RES = None
+        self.graph = None
+        self.graph_kNN = None
+        self.meld_op = None
+        self.EES = None
+        self.estimates = {}
+
+    def set_seed(self, seed):
+        """Sets random seed.
+
+        Parameters
+        ----------
+        seed : integer or numpy.RandomState
+            Random state. Defaults to the global `numpy` random number generator
+
+        Returns
+        -------
+        seed : integer or numpy.RandomState
+            Newly set random seed.
+
+        """
+
+        self.seed = seed
+        return self.seed
+
+
+    def set_phate(self, data_phate):
+        """Short summary.
+
+        Parameters
+        ----------
+        data_phate : array, shape=[n_samples, 3]
+            PHATE embedding for input data.
+
+        Returns
+        -------
+        data_phate : array, shape=[n_samples, 3]
+            Normalized PHATE embedding.
+
+        """
+        if not data_phate.shape[1] == 3:
+            raise ValueError('data_phate must have 3 dimensions')
+        if not np.isclose(data_phate.mean(), 0):
+            # data_phate must be mean-centered
+            data_phate = scipy.stats.zscore(data_phate, axis=0)
+        self.data_phate = data_phate
+
+
+    def fit_graph(self, data, **kwargs):
+        """Fits a graphtools.Graph to input data
+
+        Parameters
+        ----------
+        data : array, shape=[n_samples, n_observations]
+            Input data
+        **kwargs : dict
+            Keyword arguments passed to gt.Graph()
+
+        Returns
+        -------
+        graph : graphtools.Graph
+            Graph fit to data
+
+        """
+        self.graph = gt.Graph(data, n_pca=100, use_pygsp=True, random_state=self.seed, **kwargs)
+        return self.graph
+
+    def fit_phate(self, data, **kwargs):
+        """Generates a 3D phate embedding of input data
+
+        Parameters
+        ----------
+        data : array, shape=[n_samples, n_observations]
+            Description of parameter `data`.
+        **kwargs : dict
+            Keyword arguments passed to phate.PHATE().
+
+        Returns
+        -------
+        data_phate : array, shape=[n_samples, 3]
+            Normalized PHATE embedding for input data.
+
+        """
+        import phate
+
+        self.set_phate(phate.PHATE(n_components=3, **kwargs).fit_transform(data))
+        return self.data_phate
+
+
+    def generate_ground_truth_pdf(self, data_phate=None):
+        """Creates a random density function over input data.
+
+        Takes a set of PHATE coordinates over a set of points and creates an underlying
+        ground truth pdf over the points as a convex combination of the input phate coords.
+
+        Parameters
+        ----------
+        data_phate : array, shape=[n_samples, 3]
+            PHATE embedding for input data.
+
+        Returns
+        -------
+        pdf
+            Ground truth conditional probability of the sample given the data.
+
+        """
+
+        np.random.seed(self.seed)
+
+        if data_phate is not None:
+            self.set_phate(data_phate)
+        elif self.data_phate is None:
+            raise ValueError('data_phate must be set prior to running generate_ground_truth_pdf().')
+
+        # Create an array of values that sums to 1
+        data_simplex = np.sort(np.random.uniform(size=(2)))
+        data_simplex = np.hstack([0, data_simplex, 1])
+        data_simplex = np.diff(data_simplex)
+        np.random.shuffle(data_simplex)
+
+        # Weight each PHATE component by the simplex weights
+        sort_axis = np.sum(self.data_phate * data_simplex, axis=1)
+
+        # Pass the weighted components through a logit
+        self.pdf = scipy.special.expit(sort_axis)
+        return self.pdf
+
+    def generate_RES(self):
+        np.random.seed(self.seed)
+
+        # Create RES
+        self.RES_int = np.random.binomial(1, self.pdf)
+        self.RES = np.array(['ctrl' if res == 0 else 'expt' for res in self.RES_int])
+
+    def calculate_EES(self, data=None, **kwargs):
+        np.random.seed(self.seed)
+        if not self.graph:
+            if data is not None:
+                self.fit_graph(data)
+            else:
+                raise NameError("Must pass `data` unless graph has already been fit")
+
+        self.meld_op = meld.MELD(**kwargs, verbose=False).fit(self.graph)
+        self.EES = self.meld_op.transform(self.RES)
+        self.EES = self.EES['expt'].values # Only keep the expt condition
+        return self.EES
+
+    def calculate_mse(self, estimate):
+        '''Calculated MSE between the ground truth PDF and an estimate
+        '''
+        return sklearn.metrics.mean_squared_error(self.pdf, estimate)

meld/benchmark.py

@@ -1,3 +1,4 @@
 # Copyright (C) 2020 Krishnaswamy Lab, Yale University
 
-__version__ = "0.3.0"
+__version__ = "0.3.1"
+