From e5cfc3a4a0e71d7decd8859e908a2c80f6f9a909 Mon Sep 17 00:00:00 2001
From: James Hadfield <hadfield.james@gmail.com>
Date: Tue, 11 Apr 2023 20:56:18 +1200
Subject: [PATCH] [clades] check for multiple mutations at same pos

Multiple mutations at the same position on a single branch are now a
fatal error. Previous behaviour was to overwrite such mutations when
parsing. Suggested by #735.
---
 augur/clades.py                                | 15 +++++++++++++++
 tests/functional/clades.t                      | 11 ++++++++++-
 tests/functional/clades/toy_muts_multiple.json | 15 +++++++++++++++
 3 files changed, 40 insertions(+), 1 deletion(-)
 create mode 100644 tests/functional/clades/toy_muts_multiple.json

diff --git a/augur/clades.py b/augur/clades.py
index 30c43e8b7..ab8614d09 100644
--- a/augur/clades.py
+++ b/augur/clades.py
@@ -170,6 +170,20 @@ def is_node_in_clade(clade_alleles, node, root_sequence):
 
     return all(conditions)
 
+def ensure_no_multiple_mutations(all_muts):
+    multiples = []
+
+    for name,node in all_muts.items():
+        nt_positions = [int(mut[1:-1])-1  for mut in node.get('muts', [])]
+        if len(set(nt_positions))!=len(nt_positions):
+            multiples.append(f"Node {name} (nuc)")
+        for gene in node.get('aa_muts', {}):
+            aa_positions = [int(mut[1:-1])-1 for mut in node['aa_muts'][gene]]
+            if len(set(aa_positions))!=len(aa_positions):
+                multiples.append(f"Node {name} ({gene})")
+    
+    if multiples:
+        raise AugurError(f"Multiple mutations at the same position on a single branch were found: {', '.join(multiples)}")
 
 def assign_clades(clade_designations, all_muts, tree, ref=None):
     '''
@@ -314,6 +328,7 @@ def parse_nodes(tree_file, node_data_files):
     tree_nodes = set([clade.name for clade in tree.find_clades()])
     if not json_nodes.issubset(tree_nodes):
         raise AugurError(f"The following nodes in the node_data files ({', '.join(node_data_files)}) are not found in the tree ({tree_file}): {', '.join(json_nodes - tree_nodes)}")
+    ensure_no_multiple_mutations(node_data['nodes'])
     return (tree, node_data['nodes'])
 
 def register_parser(parent_subparsers):
diff --git a/tests/functional/clades.t b/tests/functional/clades.t
index 2b4bd759c..0dca3dc13 100644
--- a/tests/functional/clades.t
+++ b/tests/functional/clades.t
@@ -82,4 +82,13 @@ if the (branch leading to the) root has the clade-defining mutation.
 
   $ python3 "$TESTDIR/../../scripts/diff_jsons.py" clades/toy_clades_2.json "$TMP/toy_clades_2b.json" \
   >   --exclude-paths "root['generated_by']"
-  {}
\ No newline at end of file
+  {}
+
+Multiple mutations at the same position on a single branch are a fatal error
+
+  $ ${AUGUR} clades \
+  >   --tree clades/toy_tree.nwk \
+  >   --mutations clades/toy_muts_multiple.json \
+  >   --clades clades/toy_clades_nuc.tsv
+  ERROR: Multiple mutations at the same position on a single branch were found: Node A (nuc), Node AB (geneName)
+  [2]
\ No newline at end of file
diff --git a/tests/functional/clades/toy_muts_multiple.json b/tests/functional/clades/toy_muts_multiple.json
new file mode 100644
index 000000000..0688cee89
--- /dev/null
+++ b/tests/functional/clades/toy_muts_multiple.json
@@ -0,0 +1,15 @@
+{
+    "nodes": {
+      "A": {
+        "muts": ["A10T", "T10C"]
+      },
+      "AB": {
+        "aa_muts": {
+          "geneName": ["S42L", "R42H", "Y50W"]
+        }
+      },
+      "B": {
+        "muts": ["A10T", "T11C"]
+      }
+    }
+  }
\ No newline at end of file