Test t-SNE success/failure with multiple inputs

Adds tests for t-SNE behavior with multiple inputs when inputs are
formatted correctly and incorrectly. Adds a check for matching record
names in alignment and distance inputs to t-SNE when both are provided,
to make the new failure mode test pass.

src/pathogen_embed/pathogen_embed.py

@@ -309,6 +309,17 @@ def embed(args):
                 second_df = pd.DataFrame(numbers)
                 second_df.columns = ["Site " + str(k) for k in range(genomes_df.shape[1],genomes_df.shape[1] + len(numbers[0]))]
                 genomes_df = pd.concat([genomes_df, second_df], axis=1)
+
+        # If we're using PCA to initialize the t-SNE embedding, confirm that the
+        # input alignments used for PCA have the same sequence names as the
+        # input distance matrices.
+        if (
+                args.command == "t-sne" and
+                not np.array_equal(distance_matrix.index.values, np.array(sequence_names))
+        ):
+            print("ERROR: The sequence names for the distance matrix inputs do not match the names in the alignment inputs.", file=sys.stderr)
+            sys.exit(1)
+
         #performing PCA on my pandas dataframe
         pca = PCA(
             n_components=n_components,

tests/pathogen-embed-t-sne-multiple-distances-and-alignments-with-different-samples.t

@@ -0,0 +1,27 @@
+Get a distance matrix from a H3N2 HA alignment that has been sorted by sequence name.
+
+  $ pathogen-distance \
+  >   --alignment $TESTDIR/data/h3n2_ha_alignment.sorted.fasta \
+  >   --output ha_distances_complete.csv
+
+Get a distance matrix from a H3N2 NA alignment that has been sorted by sequence name.
+
+  $ pathogen-distance \
+  >   --alignment $TESTDIR/data/h3n2_na_alignment.sorted.fasta \
+  >   --output na_distances_complete.csv
+
+Remove the second record from the HA and NA distance matrices.
+This should produce mismatched records between the alignments and distances, but the pairs of alignments and distances on their own are matched.
+
+  $ cut -f 1,3- -d "," ha_distances_complete.csv | sed 2d > ha_distances.csv
+  $ cut -f 1,3- -d "," na_distances_complete.csv | sed 2d > na_distances.csv
+
+Run pathogen-embed with t-SNE on distances from H3N2 HA and H3N2 NA alignments.
+
+  $ pathogen-embed \
+  >   --alignment $TESTDIR/data/h3n2_ha_alignment.sorted.fasta $TESTDIR/data/h3n2_na_alignment.sorted.fasta \
+  >   --distance-matrix ha_distances.csv na_distances.csv \
+  >   --output-dataframe embed_t-sne.csv \
+  >   t-sne
+  ERROR: The sequence names for the distance matrix inputs do not match the names in the alignment inputs.
+  [1]

tests/pathogen-embed-t-sne-multiple-distances-and-alignments.t

@@ -0,0 +1,23 @@
+Get a distance matrix from a H3N2 HA alignment.
+
+  $ pathogen-distance \
+  >   --alignment $TESTDIR/data/h3n2_ha_alignment.fasta \
+  >   --output ha_distances.csv
+
+Get a distance matrix from a H3N2 NA alignment.
+
+  $ pathogen-distance \
+  >   --alignment $TESTDIR/data/h3n2_na_alignment.fasta \
+  >   --output na_distances.csv
+
+Run pathogen-embed with t-SNE on distances from H3N2 HA and H3N2 NA alignments.
+
+  $ pathogen-embed \
+  >   --alignment $TESTDIR/data/h3n2_ha_alignment.fasta $TESTDIR/data/h3n2_na_alignment.fasta \
+  >   --distance-matrix ha_distances.csv na_distances.csv \
+  >   --output-dataframe embed_t-sne.csv \
+  >   t-sne
+
+There should be one record in the embedding per input sequence in the alignment.
+
+  $ [[ $(sed 1d embed_t-sne.csv | wc -l) == $(grep "^>" $TESTDIR/data/h3n2_ha_alignment.fasta | wc -l) ]]