From feaf7f45456fe49285fbfbd8e7cf1d2256dde65e Mon Sep 17 00:00:00 2001
From: Samarendra <samarendrapani1998@gmail.com>
Date: Fri, 16 Aug 2024 16:54:25 +0200
Subject: [PATCH] adding an exception case to get_path

For the GRCh38 paths in the CHM13-based graph of HPRC, it does not form a path. Hence conversion now uses an exception case to not throw this warning.
---
 gaftools/cli/view.py |  5 ++---
 gaftools/gfa.py      | 23 ++++++++++++++++-------
 2 files changed, 18 insertions(+), 10 deletions(-)

diff --git a/gaftools/cli/view.py b/gaftools/cli/view.py
index 5eaefa6..3091181 100644
--- a/gaftools/cli/view.py
+++ b/gaftools/cli/view.py
@@ -65,8 +65,7 @@ def run(gaf_path, gfa=None, output=None, index=None, nodes=[], regions=[], forma
             }
             ref_contig = [contig for contig in gfa_file.contigs if gfa_file.contigs[contig] == 0]
             for contig in gfa_file.contigs:
-                contig_len[contig] = gfa_file.get_contig_length(contig)
-            print(contig_len)
+                contig_len[contig] = gfa_file.get_contig_length(contig, throw_warning=False)
             del gfa_file
         else:
             assert format == "unstable"
@@ -79,7 +78,7 @@ def run(gaf_path, gfa=None, output=None, index=None, nodes=[], regions=[], forma
             gfa_file = GFA(graph_file=gfa, low_memory=True)
             contigs = list(gfa_file.contigs.keys())
             for contig in contigs:
-                path = gfa_file.get_path(contig)
+                path = gfa_file.get_path(contig, throw_warning=False)
                 for node in path:
                     reference[contig].append(gfa_file[node])
             del gfa_file
diff --git a/gaftools/gfa.py b/gaftools/gfa.py
index 1b91a50..20a4e85 100644
--- a/gaftools/gfa.py
+++ b/gaftools/gfa.py
@@ -678,7 +678,10 @@ def return_gfa_path(self, list_of_nodes):
 
         return ",".join(path)
 
-    def get_path(self, chrom):
+    # TODO: Need to deal with non-path contigs like GRCh38-based paths in the CHM13-based minigraph rGFAs of HPRC.
+    #      The nodes have the same SN tag but do not form a path. Need to work with such cases.
+    #      Currently just create an exception case to use.
+    def get_path(self, chrom, throw_warning=True):
         """
         takes a chromosome name (matching the SN tag) and returns the path of that chromosome
         """
@@ -699,16 +702,22 @@ def get_path(self, chrom):
         if self.list_is_path(sorted_nodes):
             return sorted_nodes
         else:
-            logging.warning(
-                f"The sorted nodes with SN tag {chrom} did not create a linear path. Stopping! Returning empty list"
-            )
-            return list()
+            if throw_warning:
+                logging.warning(
+                    f"The sorted nodes with SN tag {chrom} did not create a linear path. Stopping! Returning empty list"
+                )
+                return list()
+            else:
+                logging.warning(
+                    f"The sorted nodes with SN tag {chrom} did not create a linear path. The sorted node list is returned for conversion."
+                )
+                return sorted_nodes
 
-    def get_contig_length(self, chrom):
+    def get_contig_length(self, chrom, throw_warning=True):
         """
         returns the length of the chromosome or contig name
         """
-        sorted_nodes = self.get_path(chrom)
+        sorted_nodes = self.get_path(chrom, throw_warning)
         if not sorted_nodes:
             logging.error(
                 "Was not able to return the length of the chromosome, check warning message(s)"