bugfix in drawing kdtree-based map with autoscaling

tests/test_commandline.py

@@ -17,9 +17,9 @@
 import random
 
 
-py2_hash, py3_hash = 'y7DkJVGe', 'PypAyFiw'
+py2_hash, py3_hash = 'PGiGV4YY', 'wmkqLTc4'
 data_server = 'http://dl.pasteur.fr/fop/{}/'.format(py2_hash if sys.version_info[0] == 2 else py3_hash)
-data_update = '180227'
+data_update = '180228'
 data_file = 'glycine_receptor.trxyt'
 
 data_dir = '{}_py{}_{}'.format('test_commandline', sys.version_info[0], data_update)

tramway/core/scaler.py

@@ -50,7 +50,7 @@ class Scaler(object):
 		function (callable):
 			A function that takes a data matrix as input and returns `center` and `factor`.
 			`function` is called once during the first call to :meth:`scale_point`.
-		euclidian (list):
+		euclidean (list):
 			Sequence of names or indices of the columns to be scaled by a common factor.
 	"""
 	__slots__ = ['init', 'center', 'factor', 'columns', 'function', 'euclidean']
@@ -61,7 +61,7 @@ def __init__(self, scale=None, euclidean=None):
 			scale (callable):
 				A function that takes a data matrix as input and returns `center` and 
 				`factor`. `scale` becomes the :attr:`function` attribute.
-			euclidian (list):
+			euclidean (list):
 				Sequence of names or indices of the columns to be scaled by a common 
 				factor.
 		"""
@@ -89,10 +89,10 @@ def scaled(self, points, asarray=False):
 				points = points[self.columns]
 			else:
 				if self.center is not None and isstructured(self.center):
-					raise TypeError("input data are not structured whereas scaler' is")
+					raise TypeError("input data are not structured whereas scaler's internal data are")
 				points = points[:, self.columns] 
 		elif isstructured(points):
-			raise ValueError("input data are structured whereas scaler' is not")
+			raise ValueError("input data are structured whereas scaler's internal data are not")
 		else:
 			scaler_data = self.center
 			if scaler_data is None:
@@ -153,12 +153,12 @@ def scale_point(self, points, inplace=True, scaledonly=False, asarray=False):
 			if self.function:
 				# calculate centers and factors
 				self.center, self.factor = self.function(points)
-				# equalize factor for euclidian variables
+				# equalize factor for euclidean variables
 				if self.euclidean:
 					if isinstance(points, pd.DataFrame):
 						xyz = points[self.euclidean].values
 					elif points.dtype.names:
-						xyz = np.asarray(points[self.euclidian])
+						xyz = np.asarray(points[self.euclidean])
 					else:
 						xyz = points[:,self.euclidean]
 					_, self.factor[self.euclidean] = self.function(xyz.flatten())
@@ -177,10 +177,16 @@ def scale_point(self, points, inplace=True, scaledonly=False, asarray=False):
 		if not (self.center is None and self.factor is None):
 			if not inplace:
 				points = points.copy()
-			if self.center is not None:
-				points -= self.center
-			if self.factor is not None:
-				points /= self.factor
+			if isinstance(points, np.ndarray):
+				if self.center is not None:
+					points -= np.asarray(self.center)
+				if self.factor is not None:
+					points /= np.asarray(self.factor)
+			else:
+				if self.center is not None:
+					points -= self.center
+				if self.factor is not None:
+					points /= self.factor
 		if scaledonly:
 			points = self.scaled(points, asarray)
 		elif asarray:
@@ -207,11 +213,17 @@ def unscale_point(self, points, inplace=True):
 			raise AttributeError('scaler has not been initialized')
 		if not (self.center is None and self.factor is None):
 			if not inplace:
-				points = points.copy(deep=False)
-			if self.factor is not None:
-				points *= self.factor
-			if self.center is not None:
-				points += self.center
+				points = points.copy(False)
+			if isinstance(points, np.ndarray):
+				if self.factor is not None:
+					points *= np.asarray(self.factor)
+				if self.center is not None:
+					points += np.asarray(self.center)
+			else:
+				if self.factor is not None:
+					points *= self.factor
+				if self.center is not None:
+					points += self.center
 		return points
 
 

tramway/helper/inference.py

@@ -289,6 +289,7 @@ def map_plot(maps, output_file=None, fig_format=None, \
 
 	input_file = None
 	if isinstance(maps, tuple):
+		warn('`maps` as (CellStats, str, DataFrame) tuple are deprecated', FutureWarning)
 		cells, mode, maps = maps
 	elif isinstance(maps, (pd.DataFrame, Maps)):
 		if cells is None:

tramway/helper/tessellation.py

@@ -188,10 +188,16 @@ def tessellate(xyt_data, method='gwr', output_file=None, verbose=False, \
 
 	no_nesting_error = ValueError('nesting tessellations does not apply to translocation data')
 	multiple_files = lambda a: isinstance(a, (tuple, list, frozenset, set))
-	if isinstance(xyt_data, six.string_types) or multiple_files(xyt_data):
+	xyt_files = []
+	if isinstance(xyt_data, six.string_types):
+		# file path
+		xyt_files = [xyt_data]
+	elif multiple_files(xyt_data):
 		# file path(s)
-		if multiple_files(xyt_data) and not xyt_data[1:]:
+		xyt_files = list(xyt_data)
+		if not xyt_data[1:]:
 			xyt_data = xyt_data[0]
+	if xyt_files:
 		if multiple_files(xyt_data):
 			xyt_file = xyt_data
 		else:
@@ -207,12 +213,11 @@ def tessellate(xyt_data, method='gwr', output_file=None, verbose=False, \
 					input_partition, = find_artefacts(analyses, CellStats, input_label)
 				xyt_data = analyses.data
 		if xyt_file:
-			xyt_data, xyt_path = load_xyt(xyt_data, return_paths=True, verbose=verbose)
+			xyt_data, xyt_files = load_xyt(xyt_files, return_paths=True, verbose=verbose)
 			analyses = Analyses(xyt_data)
 			if input_label is not None:
 				raise no_nesting_error
 	else:
-		xyt_path = []
 		if isinstance(xyt_data, Analyses):
 			analyses = xyt_data
 			xyt_data = analyses.data
@@ -221,7 +226,7 @@ def tessellate(xyt_data, method='gwr', output_file=None, verbose=False, \
 		#warn('TODO: test direct data input', UseCaseWarning)
 		if input_label is not None:
 			raise no_nesting_error
-	input_files = xyt_path
+	input_files = xyt_files
 
 	try:
 		setup, module = plugins[method]
@@ -439,13 +444,6 @@ def _filter_i(voronoi, cell, x):
 		stats.param['avg_distance'] = avg_distance
 	if max_distance:
 		stats.param['max_distance'] = max_distance
-	#if not plugin:
-	#	if min_location_count:
-	#		stats.param['min_location_count'] = min_location_count
-	#	if avg_location_count:
-	#		stats.param['avg_location_count'] = avg_location_count
-	#	if max_location_count:
-	#		stats.param['max_location_count'] = min_location_count
 	if knn:
 		stats.param['knn'] = knn
 	stats.param.update(kwargs)
@@ -472,9 +470,9 @@ def _filter_i(voronoi, cell, x):
 		analyses.add(Analyses(stats), label=label, comment=comment)
 
 	# save the analysis tree (`analyses`)
-	if output_file or xyt_path:
+	if output_file or xyt_files:
 		if output_file is None:
-			output_file = os.path.splitext(xyt_path[0])[0] + hdf_extensions[0]
+			output_file = os.path.splitext(xyt_files[0])[0] + hdf_extensions[0]
 
 		save_rwa(output_file, analyses, verbose, \
 			force=len(input_files)==1 and input_files[0]==output_file)

tramway/plot/map.py

@@ -46,7 +46,7 @@ def scalar_map_2d(cells, values, aspect=None, clim=None, figure=None, axes=None,
 
 	polygons = []
 	if isinstance(cells, Distributed):
-		ix, xy, ok = zip(*[ (i, c.center, 0 < c.tcount) for i, c in cells.cells.items() ])
+		ix, xy, ok = zip(*[ (i, c.center, bool(c)) for i, c in cells.items() ])
 		ix, xy, ok = np.array(ix), np.array(xy), np.array(ok)
 		voronoi = scipy.spatial.Voronoi(xy)
 		for c, r in enumerate(voronoi.point_region):
@@ -95,7 +95,13 @@ def scalar_map_2d(cells, values, aspect=None, clim=None, figure=None, axes=None,
 		except AttributeError:
 			raise TypeError('wrong type for `cells`: {}'.format(_type))
 
-	xy_min, xy_max = xy.min(axis=0), xy.max(axis=0)
+	try:
+		bounding_box = cells.descriptors(cells.bounding_box, asarray=True)
+		xy_min, xy_max = bounding_box
+	except (KeyboardInterrupt, SystemExit):
+		raise
+	except:
+		xy_min, xy_max = xy.min(axis=0), xy.max(axis=0)
 
 	scalar_map = values.loc[ix[ok]].values
 
@@ -127,6 +133,7 @@ def scalar_map_2d(cells, values, aspect=None, clim=None, figure=None, axes=None,
 		patches.set_clim(clim)
 	axes.add_collection(patches)
 
+	obj = None
 	if delaunay:
 		try:
 			import tramway.plot.mesh as mesh
@@ -136,7 +143,6 @@ def scalar_map_2d(cells, values, aspect=None, clim=None, figure=None, axes=None,
 		except:
 			import traceback
 			print(traceback.format_exc())
-			obj = None
 
 	axes.set_xlim(xy_min[0], xy_max[0])
 	axes.set_ylim(xy_min[1], xy_max[1])
@@ -149,8 +155,7 @@ def scalar_map_2d(cells, values, aspect=None, clim=None, figure=None, axes=None,
 		except AttributeError as e:
 			warn(e.args[0], RuntimeWarning)
 
-	if delaunay:
-		return obj
+	return obj
 
 
 

tramway/tessellation/base.py

@@ -833,11 +833,7 @@ def vertices(self):
 		"""Unscaled coordinates of the Voronoi vertices (numpy.ndarray)."""
 		if self._cell_centers is not None and self._vertices is None:
 			self._postprocess()
-		if isinstance(self.scaler.factor, pd.Series):
-			return np.asarray(self.scaler.unscale_point(pd.DataFrame(self._vertices, \
-				columns=self.scaler.factor.index)))
-		else:
-			return self.scaler.unscale_point(self._vertices)
+		return self.scaler.unscale_point(self._vertices, inplace=False)
 
 	@vertices.setter
 	def vertices(self, vertices):

tramway/tessellation/kdtree/__init__.py

@@ -127,7 +127,7 @@ def tessellate(self, points, **kwargs):
 		self.dichotomy.split()
 		self.dichotomy.subset = {} # clear memory
 		self.dichotomy.subset_counter = 0
-		
+
 		origin, level = zip(*[ self.dichotomy.cell[c][:2] for c in range(self.dichotomy.cell_counter) ])
 		origin = np.vstack(origin)
 		level = np.array(level)
@@ -164,7 +164,8 @@ def unique_rows(data, *args, **kwargs):
 							np.arange(j * n, (j+1) * n)[:,np.newaxis]))))
 		vertices, I = unique_rows(np.concatenate(vertices, axis=0), \
 			return_inverse=True)
-		self.vertices = vertices
+		self._vertices = vertices
+		self._lazy['vertices'] = False
 		ridge_vertices = I[np.concatenate(ridge_vertices, axis=0)]
 		u, v = ridge_vertices.T
 		nverts = vertices.shape[0]