Merge pull request #21 from blab/csv-error-check

Csv error check

src/pathogen_embed/pathogen_embed.py

@@ -196,11 +196,14 @@ def embed(args):
     if args.alignment is None and args.command == "pca":
         print("You must specify an alignment for pca, not a distance matrix", file=sys.stderr)
         sys.exit(1)
-
     # getting or creating the distance matrix
     distance_matrix = None
     if args.distance_matrix is not None:
-        distance_matrix  = pd.read_csv(args.distance_matrix, index_col=0)
+        if not args.distance_matrix.endswith('.csv'):
+            print("You must supply a CSV file for distance_matrix.", file=sys.stderr)
+            sys.exit(1)
+        else: 
+            distance_matrix  = pd.read_csv(args.distance_matrix, index_col=0)
 
     if args.alignment is not None:
         sequences_by_name = OrderedDict()
@@ -221,7 +224,11 @@ def embed(args):
     # Load embedding parameters from an external CSV file, if possible.
     external_embedding_parameters = None
     if args.embedding_parameters is not None:
-        external_embedding_parameters_df = pd.read_csv(args.embedding_parameters)
+        if not args.embedding_parameters.endswith('.csv'):
+            print("You must supply a CSV file for embedding parameters.", file=sys.stderr)
+            sys.exit(1)
+        else: 
+            external_embedding_parameters_df = pd.read_csv(args.embedding_parameters)
 
         # Get a dictionary of additional parameters provided by the external
         # file to override defaults for the current method.
@@ -421,7 +428,11 @@ def embed(args):
 
 def cluster(args):
 
-    embedding_df = pd.read_csv(args.embedding, index_col=0)
+    if not args.embedding.endswith('.csv'):
+        print("You must supply a CSV file for the embedding.", file=sys.stderr)
+        sys.exit(1)
+    else: 
+        embedding_df = pd.read_csv(args.embedding, index_col=0)
 
     clustering_parameters = {
         **({"min_cluster_size": args.min_size} if args.min_size is not None else {}),

tests/pathogen-cluster-csv-check-embedding-fail.t

@@ -0,0 +1,9 @@
+Run pathogen cluster and check that error message is thrown when non csv input is given for embedding
+
+  $ pathogen-cluster \
+  >   --embedding embed_pca.tsv \
+  >   --label-attribute pca_label \
+  >   --distance-threshold 0.5 \
+  >   --output-dataframe cluster_embed_pca.csv
+  You must supply a CSV file for the embedding.
+  [1]

tests/pathogen-embed-csv-check-dist-matrix-fail.t

@@ -0,0 +1,9 @@
+Run pathogen-embed with a tsv instead of a csv to see if it throws an error.
+
+  $ pathogen-embed \
+  >   --alignment $TESTDIR/data/h3n2_ha_alignment.fasta \
+  >   --output-dataframe embed_t-sne.tsv \
+  >   t-sne \
+  >   --perplexity 25
+You must supply a CSV file for distance_matrix.
+[1]

tests/pathogen-embed-csv-check-embedding-params-fail.t

@@ -0,0 +1,11 @@
+Run pathogen embed and check that embedding parameters with not csv input fails
+
+  $ pathogen-embed \
+  >   --alignment $TESTDIR/data/h3n2_ha_alignment.fasta \
+  >   --output-dataframe embed_t-sne.tsv \
+  >   --output-figure figure.png \
+  >   --embedding-parameters value.tsv \
+  >   t-sne \
+  >   --perplexity 25
+  You must supply a CSV file for embedding parameters.
+  [1]