Added pdbdist cmd

ariaec/commands.py

@@ -14,6 +14,7 @@ from .maplot import AriaEcContactMap
 from .econverter import AriaEcBbConverter, AriaEcXMLConverter
 from .setup import AriaEcSetup
 from .pdbqual import AriaEcPdbqual
+from .pdbdist import PDBDist
 
 
 LOG = logging.getLogger(__name__)
@@ -59,17 +60,19 @@ class AriaEcCommand(object):
     Argparse interface for aria_ec
     """
 
-    command_list = ("setup", "bbconv", "contactmap", "pdbqual", "tbl2xml")
+    command_list = ("setup", "bbconv", "contactmap", "pdbqual", "tbl2xml",
+                    "pdbdist")
     desc_list = (u"Setup ARIA infrastructure with given contacts translated "
                  u"into ARIA restraints",
                  u"Convert contacts in bbcontact format",
                  u"Contactmap tool",
                  u"Quality tool for pdb file(s)",
-                 u"XML converter for tbl distance restraint")
-    contact_types = ("evfold", "plmev", "plm", "plmdca", "plmc", "bbcontacts", "pconsc", "pconsc1",
-                     "pconsc2", "psicov", "metapsicovhb", "metapsicov_stg1",
-                     "metapsicov_stg2", "gremlin", "pdb", "native",
-                     "native_full", "contactlist")
+                 u"XML converter for tbl distance restraint",
+                 u"Extract distance distribution from culled list of pdb files")
+    contact_types = ("evfold", "plmev", "plm", "plmdca", "plmc", "bbcontacts",
+                     "pconsc", "pconsc1", "pconsc2", "psicov", "metapsicovhb",
+                     "metapsicov_stg1", "metapsicov_stg2", "gremlin", "pdb",
+                     "native", "native_full", "contactlist")
     default_confile = "conf/aria_ec.ini"
 
     def __init__(self, custom_logging=None):
@@ -249,6 +252,20 @@ class AriaEcCommand(object):
                             help="TBL distance restraint file(s)")
         return parser
 
+    @staticmethod
+    def _pdbdist_argparser(desc=None):
+        parser = argp.ArgumentParser(description=desc,
+                                     add_help=False)
+        parser.add_argument(
+            "--cullist", dest="cullpdbs", default=None,
+            metavar="CULLED_PDB_LIST", action=ReadableFile,
+            help="Culled list of pdb files")
+        parser.add_argument(
+            "--pdbdir", dest="pdbdir", default=None,
+            metavar="PDB_FOLDER",
+            help="Folder containing pdb file entries")
+        return parser
+
     def create_settings(self):
         """Create settings relative to args.command"""
 
@@ -303,7 +320,6 @@ class AriaEcCommand(object):
     def pdbqual(self):
         """
         Quality run subcommand
-        :return:
         """
         qualprot = AriaEcPdbqual(self.create_settings())
         qualprot.run()
@@ -311,19 +327,21 @@ class AriaEcCommand(object):
     def tbl2xml(self):
         """
         tbl2xml command
-
-        Returns
-        -------
-
         """
         tblconverter = AriaEcXMLConverter(self.create_settings())
         tblconverter.run_tbl2xml()
 
+    def pdbdist(self):
+        """
+        Extract pdb distance distributions
+        """
+        inst = PDBDist(self.create_settings())
+        inst.run()
+
 
 def ec2aria():
     """
     Launch ariaec command interface
-    :return:
     """
 
     mylog = CustomLogging(desc="ARIA Evolutionary Constraints Tools")

ariaec/ecsettings.py

@@ -113,9 +113,11 @@ class AriaEcSettings(Settings):
     ARIAPROJ_TEMPLATE = 'templates/aria_project_v2.3.0.xml'
     SS_DIST = 'data/ss_dist.txt'
     SCSC_MIN = 'data/scsc_min.p'
+    CULLEDPDBS = 'data/cullpdb/160427/cullpdb_pc25_res1.6_R0.25_d160427_chains3743'
     # TODO: If aria installed in site-packages, use topallhdg file from aria !
     TOPO = 'data/topallhdg5.3.pro'
-    SECTIONS = ("main", "setup", "contactmap", "bbconv", "contactdef", "pdbqual")
+    SECTIONS = ("main", "setup", "contactmap", "bbconv", "contactdef",
+                "pdbqual", "pdbdist")
 
     def __init__(self, name):
         """

ariaec/pdbdist.py

+# coding=utf-8
+"""
+                        PDB distance distribution analysis
+"""
+
+import os
+import logging
+
+from future.builtins import input
+from Bio.PDB import PDBList
+
+
+LOG = logging.getLogger(__name__)
+
+
+class PDBDist(object):
+    """
+    Exrtact pdb distance distribution
+    """
+
+    def __init__(self, settings):
+        self.settings = settings
+
+    def run(self):
+        """
+        main method
+        """
+        print(self.settings.pdbdist.args.get("pdbdir"))
+        print(self.settings.pdbdist.args.get("cullpdbs"))
+        # LOG.warning("PDB directory not given. PDBDist command has to download "
+        #             "all pdb files. This step could take at lot of time and "
+        #             "memory. Do you really want to continue ? [Y/N]")
+        # if input() not in ("y", "Y", "yes", "Yes", "YES"):
+        #     return
+        print("test")
+        pdbl = PDBList()
+        pdbl.retrieve_pdb_file(
+            '1FAT', pdir=self.settings.pdbdist.args.get('output_directory',
+                                                        os.getcwd()))
+
+        # Pour des raisons pratiques, les distances ne sont pas conservées entre
+        # chaque iteration dans la boucle
+
+        # Parcourt la liste des pdbs
+        #   Ajoute l'ensemble des distances via la fonction self.get_dist
+        pass
\ No newline at end of file

ariaec/protein.py

@@ -22,9 +22,10 @@ from .base import (reg_load, ppdict)
 LOG = logging.getLogger(__name__)
 
 
+# TODO: implement psipred format 3.5
 class SsList(object):
     """
-    Secondary structure prediction class
+    Reader for secondary prediction structure files
     """
     psipred_reg = re.compile(r'^(?P<up_index>\d+)'
                              r'\s+(?P<up_residue>[AC-IK-NP-TVWYZ])'
@@ -74,11 +75,10 @@ class SsList(object):
         self.filetype = os.path.splitext(filename)[1][1:]
         # TODO: check if given file is supported
 
-    def read(self, filename, sequence=''):
+    def read(self, filename):
         """
 
         :param filename:
-        :param sequence:
         :return:
         """
         self.check_filetype(filename)
@@ -90,11 +90,6 @@ class SsList(object):
         else:
             self.read_psipred(filename)
 
-        # TODO: since Protein.set_sec_struct call seq_sublist, should remove
-        # line below
-        # if sequence:
-        #     self.seq_sublist(sequence)
-
         LOG.debug("Secondary structure list:\n%s\n"
                   "Secondary structure dict:\n%s", self.ss_matrix,
                   self.ssdict)
@@ -140,8 +135,8 @@ class SsList(object):
         :return:
         """
         with open(filename, 'w') as psipred:
-            psipred.write("> %s\n" % desc)
-            psipred.write("".join([_[0] for _ in zip(*self.ss_matrix)[2]]))
+            psipred.write(str("> %s\n" % desc))
+            psipred.write(str("".join([_[0] for _ in zip(*self.ss_matrix)[2]])))
 
     def read_indextableplus(self, filename):
         """

ariaec/setup.py

@@ -286,7 +286,7 @@ class AriaEcSetup(object):
             outfile.write('''
 # TP number : {tp} ({ptp:.2f} %)
 # Number of contacts : {nc}
-    '''.format(tp=tp_count, ptp=ptp, nc=len(contacts)))
+'''.format(tp=tp_count, ptp=ptp, nc=len(contacts)))
 
 
 if __name__ == "__main__":

setup.py

@@ -163,6 +163,7 @@ def setup_package():
         # Project uses reStructuredText, so ensure that the docutils get
         # installed or upgraded on the target machine
         install_requires=[
+            'biopython>=1.68',
             'future>=0.15',
             'docutils>=0.3',
             'six>=1.10',