Merge pull request #92 from rigdenlab/conkit_plot-patch

Fix conkit-plot cmap

Author: FilomenoSanchez

conkit/command_line/conkit_plot.py

@@ -48,6 +48,7 @@
 
 import conkit.command_line
 import conkit.io
+from conkit.io import DISTANCE_FILE_PARSERS
 import conkit.plot
 import conkit.plot.tools
 
@@ -325,6 +326,8 @@ def main(argv=None):
 
         seq = conkit.io.read(args.seqfile, args.seqformat)[0]
         con = conkit.io.read(args.confile, args.conformat)[0]
+        if args.conformat in DISTANCE_FILE_PARSERS.keys():
+            con = con.as_contactmap()
 
         con.sequence = seq
         con.set_sequence_register()
@@ -335,6 +338,8 @@ def main(argv=None):
 
         if args.otherfile:
             other = conkit.io.read(args.otherfile, args.otherformat)[0]
+            if args.otherformat in DISTANCE_FILE_PARSERS.keys():
+                other = other.as_contactmap()
             other.sequence = seq
             other.set_sequence_register()
             other.remove_neighbors(min_distance=args.dtn, inplace=True)
@@ -344,15 +349,18 @@ def main(argv=None):
             other_sliced = None
 
         if args.reffile:
+            if args.refformat not in ["pdb", "mmcif"]:
+                msg = "The provided format {0} is not yet implemented for the reference flag".format(args.refformat)
+                raise RuntimeError(msg)
+
             if args.refid:
                 reference = conkit.io.read(args.reffile, args.refformat)[args.refid]
             else:
                 reference = conkit.io.read(args.reffile, args.refformat)[0]
 
-            if args.refformat not in ["pdb", "mmcif"]:
-                msg = "The provided format {0} is not yet implemented for the reference flag".format(args.refformat)
-                raise RuntimeError(msg)
-
+            reference = reference.as_contactmap()
+            reference.sequence = seq
+            reference.set_sequence_register()
             con_matched = con_sliced.match(reference, match_other=True, renumber=True, remove_unmatched=True)
             if other_sliced:
                 other_matched = other_sliced.match(reference, renumber=True, remove_unmatched=True)
@@ -393,6 +401,8 @@ def altloc_remove(cmap):
 
         seq = conkit.io.read(args.seqfile, args.seqformat)[0]
         con = conkit.io.read(args.confile, args.conformat)[0]
+        if args.conformat in DISTANCE_FILE_PARSERS:
+            con = con.as_contactmap()
 
         con.sequence = seq
         con.set_sequence_register()
@@ -407,9 +417,13 @@ def altloc_remove(cmap):
     elif args.which == "covariance_validation":
         if args.pdbformat != 'pdb':
             raise ValueError('Model file format can only be PDB')
+        elif args.distformat not in DISTANCE_FILE_PARSERS.keys():
+            raise ValueError('Prediction file format can only be a distance file')
+
         sequence = conkit.io.read(args.seqfile, args.seqformat)[0]
         if len(sequence) < 5:
             raise ValueError('Cannot validate a model with less than 5 residues')
+
         prediction = conkit.io.read(args.distfile, args.distformat)[0]
         model = conkit.io.read(args.pdbfile, args.pdbformat)[0]
         p = PDBParser()
@@ -420,12 +434,17 @@ def altloc_remove(cmap):
         figure_aspect_ratio = None
 
     elif args.which == "distogram_heatmap":
+        if args.distformat not in DISTANCE_FILE_PARSERS.keys():
+            raise ValueError('Prediction file format can only be a distance file')
+
         sequence = conkit.io.read(args.seqfile, args.seqformat)[0]
         distogram = conkit.io.read(args.distfile, args.distformat)[0]
         distogram.sequence = sequence
         distogram.set_sequence_register()
 
         if args.otherfile:
+            if args.otherformat not in DISTANCE_FILE_PARSERS.keys():
+                raise ValueError('Other prediction file format can only be a distance file')
             other = conkit.io.read(args.otherfile, args.otherformat)[0]
             other.sequence = sequence
             other.set_sequence_register()
@@ -439,12 +458,16 @@ def altloc_remove(cmap):
 
         seq = conkit.io.read(args.seqfile, args.seqformat)[0]
         con = conkit.io.read(args.confile, args.conformat)[0]
+        if args.conformat in DISTANCE_FILE_PARSERS:
+            con = con.as_contactmap()
 
         con.sequence = seq
         con.set_sequence_register()
 
         if args.otherfile:
             other = conkit.io.read(args.otherfile, args.otherformat)[0]
+            if args.otherformat in DISTANCE_FILE_PARSERS:
+                other = other.as_contactmap()
             other.sequence = seq
             other.set_sequence_register()
         else:
@@ -460,6 +483,8 @@ def altloc_remove(cmap):
 
         seq = conkit.io.read(args.seqfile, args.seqformat)[0]
         con = conkit.io.read(args.confile, args.conformat)[0]
+        if args.conformat in DISTANCE_FILE_PARSERS:
+            con = con.as_contactmap()
 
         con.sequence = seq
         con.set_sequence_register()
@@ -482,6 +507,8 @@ def altloc_remove(cmap):
 
         seq = conkit.io.read(args.seqfile, args.seqformat)[0]
         con = conkit.io.read(args.confile, args.conformat)[0]
+        if args.conformat in DISTANCE_FILE_PARSERS:
+            con = con.as_contactmap()
 
         con.sequence = seq
         con.set_sequence_register()
@@ -492,6 +519,8 @@ def altloc_remove(cmap):
             pdb = conkit.io.read(args.pdbfile, "pdb")[args.pdbchain]
         else:
             pdb = conkit.io.read(args.pdbfile, "pdb")[0]
+
+        pdb = pdb.as_contactmap()
         con_matched = con.match(pdb, renumber=True, remove_unmatched=True)
 
         figure = conkit.plot.PrecisionEvaluationFigure(

conkit/core/distance.py

@@ -62,7 +62,23 @@ class Distance(Contact):
 
     """
 
-    __slots__ = ["distance_bins", "distance_scores"]
+    __slots__ = [
+        "_distance_bound",
+        "raw_score",
+        "_res1",
+        "_res2",
+        "res1_chain",
+        "res2_chain",
+        "_res1_seq",
+        "_res2_seq",
+        "_res1_altseq",
+        "_res2_altseq",
+        "scalar_score",
+        "_status",
+        "weight",
+        "distance_bins",
+        "distance_scores"
+    ]
 
     def __init__(self, res1_seq, res2_seq, distance_scores, distance_bins, raw_score=None, distance_bound=(0, 8)):
         """Initialize a generic distance residue pair
@@ -202,6 +218,29 @@ def _reshape_bins(self, new_bins):
         self.distance_bins = tuple(new_bins)
         self.distance_scores = tuple(new_distance_scores)
 
+    def as_contact(self, distance_cutoff):
+        """Create a :obj:`~conkit.core.contact.Contact` instance with the information in this
+        :obj:`~conkit.core.distance.Distance` instance.
+
+        Parameters
+        ----------
+        distance_cutoff : int, float
+           The distance cutoff used to consider a residue pair within contact of each other
+
+        Returns
+        -------
+        :obj:`~conkit.core.contact.Contact`
+            A contact with the information present in this distance instance.
+        """
+        if self.predicted_distance > distance_cutoff:
+            contact = Contact(self.res1_seq, self.res2_seq, 0, distance_bound=(0, distance_cutoff))
+        else:
+            contact = Contact(self.res1_seq, self.res2_seq, self.raw_score, distance_bound=(0, distance_cutoff))
+        for attr in self.__slots__:
+            if hasattr(contact, attr):
+                setattr(contact, attr, getattr(self, attr))
+        return contact
+
     @staticmethod
     def _assert_valid_bins(distance_bins):
         """Determine whether a set of distance bins is valid. Valid distance bins must follow these rules:

conkit/core/distancefile.py

@@ -65,7 +65,7 @@ class DistanceFile(ContactFile):
 
     """
 
-    __slots__ = ["_original_file_format"]
+    __slots__ = ["author", "target", "_method", "_remark", "_original_file_format"]
 
     def __init__(self, id):
         """Initialise a new distance file"""

conkit/core/distogram.py

@@ -64,7 +64,7 @@ class Distogram(ContactMap):
 
     """
 
-    __slots__ = ["_original_file_format"]
+    __slots__ = ["_original_file_format", "_sequence"]
 
     def __init__(self, id):
         self._original_file_format = None
@@ -228,8 +228,10 @@ def as_contactmap(self, distance_cutoff=8):
         contactmap = ContactMap("map_1")
         for dist in self:
             if dist.predicted_distance <= distance_cutoff:
-                contact = Contact(dist.res1_seq, dist.res2_seq, dist.raw_score, distance_bound=(0, distance_cutoff))
+                contact = dist.as_contact(distance_cutoff)
                 contactmap.add(contact)
+        if self.sequence is not None:
+            contactmap.sequence = self.sequence._inplace(False)
 
         return contactmap
 

conkit/plot/contactmap.py

@@ -286,4 +286,4 @@ def color_selector(contact):
             else:
                 return ColorDefinitions.GENERAL
 
-        return [color_selector(contact) for contact in h]
+        return [color_selector(contact) for contact in h if Gap.IDENTIFIER not in (contact.res1_seq, contact.res2_seq)]

docs/examples/code/plot_map_reference.py

@@ -39,6 +39,7 @@
 # Now we need to compare it to our reference structure
 pdb_file = "toxd/toxd.pdb"
 pdb = conkit.io.read(pdb_file, "pdb").top
+pdb = pdb.as_contactmap()
 # The two keywords do the following:
 #       - match_other      : renumber the pdb to match gaps in target
 #       - remove_unmatched : remove contacts absent from the pdb_file

docs/examples/rst/python_analyse_conpred.rst

@@ -65,6 +65,7 @@ In ConKit, this functionality is provided via the :attr:`~conkit.core.contactmap
 .. code-block:: python
 
    >>> pdb = conkit.io.read("toxd/toxd.pdb", "pdb").top_map
+   >>> pdb = pdb.as_contactmap()
    >>> conpred.remove_neighbors(inplace=True)
    >>> conpred.sort("raw_score", reverse=True, inplace=True)
    >>> conpred.match(pdb, inplace=True)