[translate] require nodes to have sequences

This is the implicit expectation, and is true for all of our pipelines.
In theory it's possible to translate without having the full sequence
attached to the node by reconstructing mutations on the root, but that
is not the approach taken by the current code.

As we explicitly check the tree nodes against the sequences in the
node-data JSON we can skip the automatic tests optionally performed when
reading the node-data JSON.

Closes #1345

augur/translate.py

@@ -20,6 +20,7 @@
 from .utils import read_node_data, load_features, write_json, get_json_name
 from treetime.vcf_utils import read_vcf
 from augur.errors import AugurError
+from textwrap import dedent
 
 class MissingNodeError(Exception):
     pass
@@ -341,14 +342,22 @@ def sequences_json(node_data_json, tree):
     Extract the full nuc sequence for each node in the provided node-data JSON.
     Returns a dict, keys are node names and values are a string of the genome sequence (nuc)
     """
-    node_data = read_node_data(node_data_json, tree)
+    node_data = read_node_data(node_data_json)
     if node_data is None:
         raise AugurError("could not read node data (incl sequences)")
     # extract sequences from node meta data
     sequences = {}
     for k,v in node_data['nodes'].items():
         if 'sequence' in v:
             sequences[k] = v['sequence']
+    tree_nodes = {c.name for c in tree.find_clades()}
+    tree_nodes_missing_sequences = tree_nodes - set(sequences.keys())
+    if len(tree_nodes_missing_sequences):
+        raise AugurError(dedent(f"""\
+            {len(tree_nodes_missing_sequences)} nodes on the tree are missing nucleotide sequences in the node-data JSON.
+            These must be present under 'nodes' → <node_name> → 'sequence'.
+            This error may originate from using 'augur ancestral' with VCF input; in this case try using VCF output from that command here.
+            """))
     return sequences
 
 def register_parser(parent_subparsers):
@@ -414,7 +423,7 @@ def run(args):
         if len(features_without_variation):
             print("{} genes had no mutations and so have been be excluded.".format(len(features_without_variation)))  
     else:
-        sequences = sequences_json(args.ancestral_sequences, args.tree)
+        sequences = sequences_json(args.ancestral_sequences, tree)
         translations = {fname: translate_feature(sequences, feat) for fname, feat in features.items() if feat.type != 'source'}
 
     ## glob the annotations for later auspice export