Implemented input file detection for visualize module

radseq_analysis/modules/visualize.py

 from radseq_analysis import visualization
 
 
-def detect_plot_type(input_file):
-    return 'frequencies'
+def detect_plot_type(input_file_path):
+
+    input_file = open(input_file_path)
+    line = input_file.readline()
+    if line == 'Frequency\tCount\n':
+        return 'frequencies'
+    elif line.startswith('Stack_ID\tHaplotype_ID\tSequence\tMatches\tMismatches\tGaps'):
+        return 'clustering'
+    else:
+        temp = set()
+        fields = line[:-1].split('\t')
+        temp.add(len(fields))
+        loci_matrix = True
+        for f in fields:
+            try:
+                int(f)
+            except ValueError:
+                loci_matrix = False
+        for line in input_file:
+            if line:
+                fields = line[:-1].split('\t')
+                temp.add(len(fields))
+                for f in fields:
+                    try:
+                        int(f)
+                    except ValueError:
+                        loci_matrix = False
+        if loci_matrix and len(temp) == 1:
+            return 'haplotypes'
+        else:
+            return 'error'
 
 
 def analysis(input_file_path, popmap_file_path, output_file_path, parameters):
@@ -10,13 +39,21 @@ def analysis(input_file_path, popmap_file_path, output_file_path, parameters):
     plot_type = detect_plot_type(input_file_path)
 
     if plot_type == 'haplotypes':
+        print('Generating plot from haplotypes matrix')
         visualization.haplotypes(input_file_path,
                                  output_file_path,
                                  parameters.species)
     elif plot_type == 'clustering':
+        print('Generating plot from alleles file for clustering')
         visualization.clustering(input_file_path,
                                  popmap_file_path,
                                  output_file_path)
     elif plot_type == 'frequencies':
+        print('Generating plot from frequencies file')
         visualization.frequencies(input_file_path,
                                   output_file_path)
+    elif plot_type == 'error':
+        print('\nError: could not detect input file type. Input file should be :\n' +
+              '\t- a haplotypes matrix generated by the heatmap command\n' +
+              '\t- an allele file generated by the rescue command\n' +
+              '\t- a frequencies file generated by the frequencies command\n')