CoordinateSource patches (#582)

* take set of mrc sizes (this was the original intention)

* add custom number-parsable method

* accomodate floats in coordinate files

* add tests covering floats in coord files

Author: chris-langfield

src/aspire/source/coordinates.py

@@ -95,7 +95,7 @@ def __init__(self, files, particle_size, max_rows):
             f"{self.__class__.__name__} from {os.path.dirname(self.mrc_paths[0])} contains {len(mrc_paths)} micrographs, {len(self.particles)} picked particles."
         )
         # report different mrc shapes
-        logger.info(f"Micrographs have the following shapes: {*self.mrc_shapes,}")
+        logger.info(f"Micrographs have the following shapes: {*set(self.mrc_shapes),}")
 
         # remove particles whose boxes do not fit at given particle_size
         # and get number removed
@@ -325,6 +325,14 @@ def _images(self, start=0, num=np.inf, indices=None):
 
         return Image(im)
 
+    @staticmethod
+    def _is_number(text):
+        """
+        Used in validation of coordinate files. We allow strings containing
+        - or . to account for negative values and floats.
+        """
+        return text.replace("-", "1").replace(".", "1").isdigit()
+
 
 class BoxesCoordinateSource(CoordinateSource):
     """
@@ -355,7 +363,7 @@ def _extract_box_size(self, box_file):
         with open(box_file, "r") as box:
             first_line = box.readlines()[0].split()
             if len(first_line) >= 4:
-                box_size = int(first_line[2])  # x size or y size works
+                box_size = int(float(first_line[2]))  # x size or y size works
                 return box_size
             else:
                 logger.error(f"Problem with coordinate file: {box_file}")
@@ -380,14 +388,14 @@ def _validate_box_file(self, box_file, global_particle_size):
                         "flag in aspire extract-particles."
                     )
 
-                if not all(p.isnumeric() for p in line.split()):
+                if not all(self._is_number(p) for p in line.split()):
                     logger.error(f"Problem with coordinate file: {box_file}")
                     raise ValueError(
                         "Coordinate file contains non-numeric coordinate values."
                     )
 
                 # we can only accept square particles
-                size_x, size_y = int(line.split()[2]), int(line.split()[3])
+                size_x, size_y = float(line.split()[2]), float(line.split()[3])
                 if size_x != size_y:
                     logger.error(f"Problem with coordinate file: {box_file}")
                     raise ValueError(
@@ -429,7 +437,7 @@ def _coords_list_from_file(self, coord_file):
         with open(coord_file, "r") as infile:
             lines = [line.split() for line in infile.readlines()]
         # coords are already in box format, so simply cast to int
-        return [[int(x) for x in line] for line in lines]
+        return [[int(float(x)) for x in line] for line in lines]
 
     def _force_new_particle_size(self, new_size):
         """
@@ -485,7 +493,7 @@ def _validate_centers_file(self, coord_file):
                         "Coordinate file contains a line with less than 2 numbers."
                     )
                 # check that the coordinate has numeric values
-                if not all(c.isnumeric() for c in line.split()):
+                if not all(self._is_number(c) for c in line.split()):
                     logger.error(f"Problem with coordinate file: {coord_file}")
                     raise ValueError(
                         "Coordinate file contains non-numeric coordinate values."
@@ -504,7 +512,7 @@ def _validate_starfile(self, coord_file):
             )
         # check that all values in each column are numeric
         if not all(
-            all(df[col].str.isnumeric())
+            all(df[col].apply(self._is_number))
             for col in ["_rlnCoordinateX", "_rlnCoordinateY"]
         ):
             logger.error(f"Problem with coordinate file: {coord_file}")
@@ -531,5 +539,5 @@ def _coords_list_from_file(self, coord_file):
             return self._coords_list_from_star(coord_file)
         # otherwise we assume text file format with one coord per line:
         with open(coord_file, "r") as infile:
-            lines = [line.split() for line in infile.readlines()]
+            lines = [[float(c) for c in line.split()] for line in infile.readlines()]
         return [self._box_coord_from_center(line, self.particle_size) for line in lines]

tests/test_coordinate_source.py

@@ -85,6 +85,13 @@ def setUp(self):
             self.createTestCoordFiles(_centers, i)
             self.createTestStarFiles(_centers, i)
 
+        # Create extra coordinate files with float
+        # coordinates to make sure we can process these
+        # as well
+        self.createFloatBoxFile(centers)
+        self.createFloatCoordFile(centers)
+        self.createFloatStarFile(centers)
+
         # create lists of files
         self.all_mrc_paths = sorted(glob(os.path.join(self.data_folder, "sample*.mrc")))
         # create file lists that will be used several times
@@ -113,6 +120,10 @@ def setUp(self):
             )
         )
 
+        self.float_box = os.path.join(self.data_folder, "float.box")
+        self.float_coord = os.path.join(self.data_folder, "float.coord")
+        self.float_star = os.path.join(self.data_folder, "float.star")
+
     def tearDown(self):
         self.tmpdir.cleanup()
 
@@ -166,6 +177,47 @@ def createTestStarFiles(self, centers, index):
         starfile = StarFile(blocks=blocks)
         starfile.write(star_fp)
 
+    def createFloatBoxFile(self, centers):
+        # for testing float coordinates
+        # create a box file (lower left corner and X/Y sizes)
+        box_fp = os.path.join(self.data_folder, "float.box")
+        # populate box file with coordinates in box format
+        with open(box_fp, "w") as box:
+            for center in centers:
+                # to make a box file, we convert the centers to lower left
+                # corners by subtracting half the particle size (here: 256)
+                lower_left_corners = (center[0] - 128, center[1] - 128)
+                box.write(
+                    f"{lower_left_corners[0]}.000\t{lower_left_corners[1]}.000\t256.000\t256.000\n"
+                )
+
+    def createFloatCoordFile(self, centers):
+        # for testing float coordinates
+        # create a coord file (only particle centers listed)
+        coord_fp = os.path.join(self.data_folder, "float.coord")
+        # populate coord file with particle centers
+        with open(coord_fp, "w") as coord:
+            for center in centers:
+                # .coord file usually contains just the centers
+                coord.write(f"{center[0]}.000\t{center[1]}.000\n")
+
+    def createFloatStarFile(self, centers):
+        # for testing float coordinates
+        # create a star file (only particle centers listed)
+        star_fp = os.path.join(self.data_folder, "float.star")
+        # populate star file with particle centers
+        x_coords = [str(center[0]) + ".000" for center in centers]
+        y_coords = [str(center[1]) + ".000" for center in centers]
+        blocks = OrderedDict(
+            {
+                "coordinates": DataFrame(
+                    {"_rlnCoordinateX": x_coords, "_rlnCoordinateY": y_coords}
+                )
+            }
+        )
+        starfile = StarFile(blocks=blocks)
+        starfile.write(star_fp)
+
     def testLoadFromBox(self):
         # ensure successful loading from box files
         BoxesCoordinateSource(self.files_box)
@@ -178,6 +230,24 @@ def testLoadFromStar(self):
         # ensure successful loading from particle center files (.star)
         CentersCoordinateSource(self.files_star, particle_size=256)
 
+    def testLoadFromBox_Floats(self):
+        # ensure successful loading from box files with float coordinates
+        BoxesCoordinateSource([(self.all_mrc_paths[0], self.float_box)])
+
+    def testLoadFromCenters_Floats(self):
+        # ensure successful loading from particle center files (.coord)
+        # with float coordinates
+        CentersCoordinateSource(
+            [(self.all_mrc_paths[0], self.float_coord)], particle_size=256
+        )
+
+    def testLoadFromStar_Floats(self):
+        # ensure successful loading from particle center files (.star)
+        # with float coordinates
+        CentersCoordinateSource(
+            [(self.all_mrc_paths[0], self.float_star)], particle_size=256
+        )
+
     def testNonSquareParticles(self):
         # nonsquare box sizes must fail
         with self.assertRaises(ValueError):