Added a new GODag function which creates a hierarchy report for all t…

…erms below a user specified GO Term.

If the user does not specify a GO Term, then a hierarchy report is generated for each: BP, MF, CC

goatools/obo_parser.py

@@ -162,6 +162,22 @@ def get_all_child_edges(self):
             all_child_edges |= p.get_all_child_edges()
         return all_child_edges
 
+    def write_hier_rec(self, out=sys.stdout, 
+                      len_dash=1, max_depth=None, num_child=None,
+                      depth=1, dp="-"):
+        """Write hierarchy for a GO Term record."""
+        if len_dash is not None:
+          dp = ''.join(['-']*depth) if len_dash is not None else ''
+          out.write('{DASHES:{N}} '.format(DASHES=dp, N=len_dash))
+        if num_child is not None:
+          out.write('{N:>5} '.format(N=len(self.get_all_children())))
+        out.write('{GO}\n'.format(GO=self))
+        depth += 1
+        if max_depth is not None and depth > max_depth:
+          return
+        for p in self.children:
+            p.write_hier_rec(out, len_dash, max_depth, num_child, depth, dp)
+
 
 class GODag(dict):
 
@@ -206,7 +222,8 @@ def _init_depth(rec):
         # populate children and levels
         for rec in self.values():
             for p in rec.parents:
-                p.children.append(rec)
+                if rec not in p.children:
+                    p.children.append(rec)
 
             if rec.level is None:
                 _init_level(rec)
@@ -215,9 +232,21 @@ def _init_depth(rec):
                 _init_depth(rec)
 
     def write_dag(self, out=sys.stdout):
-        """Write info for all GO Term in obo file, sorted numerically."""
+        """Write info for all GO Terms in obo file, sorted numerically."""
         for rec_id, rec in sorted(self.items()):
             print(rec, file=out)
+
+    def write_hier_all(self, out=sys.stdout, 
+                      len_dash=1, max_depth=None, num_child=None):
+        """Write hierarchy for all GO Terms in obo file."""
+        # Print: [biological_process, molecular_function, and cellular_component]
+        for go_id in ['GO:0008150', 'GO:0003674', 'GO:0005575']:
+          self.write_hier(go_id, out, len_dash, max_depth, num_child) 
+
+    def write_hier(self, GO_id, out=sys.stdout, 
+                       len_dash=1, max_depth=None, num_child=None):
+        """Write hierarchy for a GO Term."""
+        self[GO_id].write_hier_rec(out, len_dash, max_depth, num_child)
 
     def write_summary_cnts(self, GO_ids, out=sys.stdout):
         """Write summary of level and depth counts for specific GO ids."""

scripts/write_hierarchy.py

@@ -0,0 +1,92 @@
+#!/usr/bin/env python
+# -*- coding: UTF-8 -*-
+from __future__ import print_function
+
+"""
+Print the hierarchy below Term, GO:0030663
+>>> python {SCR} GO:0030663
+
+- GO:0030663	level-05	depth-07	COPI-coated vesicle membrane [cellular_component]
+-- GO:0012508	level-05	depth-08	Golgi to ER transport vesicle membrane [cellular_component]
+-- GO:0012509	level-05	depth-08	inter-Golgi transport vesicle membrane [cellular_component]
+
+
+Write the hierarchy below Term, GO:0030663 into a file
+>>> python {SCR} GO:0030663 --o=hier_GO_0030663.rpt
+
+  WROTE: hier_GO_0030663.rpt
+
+Print the hierarchy for biological process, molecular_function, and cellular_component:
+>>> python {SCR} --o=hier_BP_MF_CC.rpt
+
+Print hierarchy for BP, MF, CC only printing the first 2 levels.
+>>> python {SCR} --max_depth=2
+>>> python {SCR} --max_depth=2 --dash_len=2 --num_child
+
+Print hierarchy
+-  26894 GO:0008150	level-00	depth-00	biological_process [biological_process]
+--    30 GO:0001906	level-01	depth-01	cell killing [biological_process]
+--   555 GO:0002376	level-01	depth-01	immune system process [biological_process]
+-- 11208 GO:0065007	level-01	depth-01	biological regulation [biological_process]
+
+>>> python {SCR} 
+
+This program prints the hierarchy for all GO terms, if no argument is provided.
+If a GO term is provided as an argument, then the hierarchy of all children 
+for that term is printed.
+
+""".format(SCR=__file__)
+
+import sys
+import os
+sys.path.insert(0, os.path.join(os.path.dirname(__file__), ".."))
+from goatools.obo_parser import GODag
+
+
+if __name__ == "__main__":
+
+    import argparse
+    p = argparse.ArgumentParser(__doc__,
+                 formatter_class=argparse.ArgumentDefaultsHelpFormatter)
+
+    p.add_argument('go_ids', type=str, nargs='*',
+                 help='GO Term, e.g. GO:0070458')
+    p.add_argument('--o', default=None, type=str,
+                 help="Specifies the name of the output file")
+    p.add_argument('--no_indent', default=False,
+                 help="Do not indent GO terms", action='store_true')
+    p.add_argument('--obo', default="go-basic.obo", type=str,
+                 help="Location and name of the obo file")
+    p.add_argument('--dash_len', default=1, type=int,
+                 help="Printed width of the dashes column")
+    p.add_argument('--max_depth', default=None, type=int,
+                 help="max depth for printing relative to GO Term")
+    p.add_argument('--num_child', default=None, action='store_true',
+                 help="Print total number of children for each GO")
+
+    args = p.parse_args()
+
+    obo_dag = GODag(obo_file=args.obo)
+
+    file_out = sys.stdout if args.o is None else open(args.o, 'w')
+    lenprt = args.dash_len if not args.no_indent else None
+
+    if args.go_ids:
+      for go_id in args.go_ids:
+        obo_dag.write_hier(
+            go_id, 
+            file_out, 
+            len_dash=lenprt,
+            max_depth=args.max_depth,
+            num_child=args.num_child)
+    else:
+      obo_dag.write_hier_all(
+          file_out, 
+          len_dash=lenprt,
+          max_depth=args.max_depth,
+          num_child=args.num_child)
+
+    if args.o is not None:
+      file_out.close()
+      sys.stdout.write("  WROTE: {}\n".format(args.o))
+