Added the ability to compress/shorten the hierarchy reports by not

printing a path to a GO leaf-level term more than once.

goatools/obo_parser.py

@@ -162,21 +162,31 @@ 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,
+    def write_hier_rec(self, gos_printed, out=sys.stdout, 
+                      len_dash=1, max_depth=None, num_child=None, short_prt=False,
                       depth=1, dp="-"):
         """Write hierarchy for a GO Term record."""
+        GO_id = self.id
+        nrp = short_prt and GO_id in gos_printed
         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))
+            letter = '-' if not nrp or not self.children else '='
+            dp = ''.join([letter]*depth)
+            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('{N:>5} '.format(N=len(self.get_all_children())))
         out.write('{GO}\n'.format(GO=self))
+        # Track GOs previously printed only if needed
+        if short_prt:
+          gos_printed.add(GO_id)
+        # Do not print hierarchy below this turn if it has already been printed
+        if nrp:
+            return 
         depth += 1
         if max_depth is not None and depth > max_depth:
-          return
+            return
         for p in self.children:
-            p.write_hier_rec(out, len_dash, max_depth, num_child, depth, dp)
+            p.write_hier_rec(gos_printed, out, len_dash, max_depth, num_child, short_prt, 
+                depth, dp)
 
 
 class GODag(dict):
@@ -237,16 +247,17 @@ def write_dag(self, out=sys.stdout):
             print(rec, file=out)
 
     def write_hier_all(self, out=sys.stdout, 
-                      len_dash=1, max_depth=None, num_child=None):
+                      len_dash=1, max_depth=None, num_child=None, short_prt=False):
         """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) 
+          self.write_hier(go_id, out, len_dash, max_depth, num_child, short_prt) 
 
     def write_hier(self, GO_id, out=sys.stdout, 
-                       len_dash=1, max_depth=None, num_child=None):
+                       len_dash=1, max_depth=None, num_child=None, short_prt=False):
         """Write hierarchy for a GO Term."""
-        self[GO_id].write_hier_rec(out, len_dash, max_depth, num_child)
+        gos_printed = set()
+        self[GO_id].write_hier_rec(gos_printed, out, len_dash, max_depth, num_child, short_prt)
 
     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

@@ -23,6 +23,19 @@
 >>> python {SCR} --max_depth=2
 >>> python {SCR} --max_depth=2 --dash_len=2 --num_child
 
+
+Print a shortened version of the hierarchy for BP, MF, and CC.
+This will only print a path to a leaf GO Term once.  
+If the path appears a second time, the term is printed again, but its path is not.
+The presence of a compressed (unprinted) paths is marked by using '=" instead of '-'.
+
+    $ wc -l hier_BP_MF_CC*.rpt
+
+          789583 hier_BP_MF_CC.rpt
+           70152 hier_BP_MF_CC_short.rpt
+
+>>> python {SCR} --o=hier_BP_MF_CC_short.rpt --short
+
 Print hierarchy
 -  26894 GO:0008150	level-00	depth-00	biological_process [biological_process]
 --    30 GO:0001906	level-01	depth-01	cell killing [biological_process]
@@ -62,7 +75,10 @@
     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")
+                 help="Print count of total number of children for each GO")
+    p.add_argument('--short', default=False, action='store_true',
+                 help="If a branch has already been printed, do not re-print." 
+                      "Print '===' instead of dashes to note the point of compression")
 
     args = p.parse_args()
 
@@ -78,13 +94,15 @@
             file_out, 
             len_dash=lenprt,
             max_depth=args.max_depth,
-            num_child=args.num_child)
+            num_child=args.num_child,
+            short_prt=args.short)
     else:
       obo_dag.write_hier_all(
           file_out, 
           len_dash=lenprt,
           max_depth=args.max_depth,
-          num_child=args.num_child)
+          num_child=args.num_child,
+          short_prt=args.short)
 
     if args.o is not None:
       file_out.close()