Updates saving of output files for querying

ggCaller/__main__.py

@@ -59,9 +59,9 @@ def get_options():
                     default=False,
                     help='Save graph objects for sequence querying. '
                          '[Default = False] ')
-    IO.add_argument('--data',
+    IO.add_argument('--prev-run',
                     default=None,
-                    help='Directory containing data from previous ggCaller run generated via "--save" ')
+                    help='Directory containing data from previous ggCaller run. Must have been run with "--save" ')
     IO.add_argument(
         "--all-seq-in-graph",
         dest="all_seq_in_graph",
@@ -400,6 +400,9 @@ def main():
     Path_dir = os.path.join(output_dir, "Path_dir")
     if not os.path.exists(Path_dir):
         os.mkdir(Path_dir)    
+
+    # ensure trailing slash present
+    Path_dir = os.path.join(Path_dir, "")
 
     # if build graph specified, build graph and then call ORFs
     if (options.graph is not None) and (options.colours is not None) and (options.query is None):
@@ -408,10 +411,10 @@ def main():
     # query unitigs in previous saved ggc graph
     elif (options.graph is not None) and (options.colours is not None) and (options.refs is None) and \
             (options.query is not None):
-        if options.data is None:
-            print("Please specify a ggc_data directory from a previous ggCaller run.")
+        if options.prev_run is None:
+            print("Please specify a directory from a previous ggCaller run.")
             sys.exit(1)
-        search_graph(graph, options.graph, options.colours, options.query, options.data, output_dir, options.query_id,
+        search_graph(graph, options.graph, options.colours, options.query, options.prev_run, output_dir, options.query_id,
                      options.threads)
         print("Finished.")
         sys.exit(0)
@@ -502,6 +505,12 @@ def main():
     ORFMap_dir = os.path.join(output_dir, "ORF_dir")
     if not os.path.exists(ORFMap_dir):
         os.mkdir(ORFMap_dir)
+
+    # ensure trailing slash present
+    ORFMap_dir = os.path.join(ORFMap_dir, "")
+
+    # save the kmer_array to ORFMap_dir
+    graph[0].data_out(ORFMap_dir + "kmer_array.dat")
 
     # load models models if required
     if not options.no_filter:
@@ -552,9 +561,6 @@ def main():
             # make sure trailing forward slash is present
             objects_dir = os.path.join(objects_dir, "")
 
-            # serialise graph object and high scoring ORFs to future reading
-            graph[0].data_out(objects_dir + "ggc_graph.dat")
-
             # create index of all high_scoring_ORFs node_IDs
             node_index = defaultdict(list)
             for colour_ID, file_path in ORF_file_paths.items():

src/graph.cpp

@@ -314,7 +314,7 @@ std::pair<std::map<size_t, std::string>, std::map<size_t, std::string>> Graph::f
                 {
                     const std::string base_filename = FM_fasta_file.substr(FM_fasta_file.find_last_of("/\\") + 1);
 
-                    const auto idx_file_name = path_dir + "/" + base_filename + ".fmp";
+                    const auto idx_file_name = path_dir + base_filename + ".fmp";
                     if (!load_from_file(fm_idx, idx_file_name))
                     {
                         cout << "FM-Index not available for " << FM_fasta_file << endl;
@@ -338,7 +338,7 @@ std::pair<std::map<size_t, std::string>, std::map<size_t, std::string>> Graph::f
 
             // write ORF_map file
             {
-                std::string ORF_file_path = tmp_dir + "/colour_" + std::to_string(colour_ID) + "_ORFs.tmp";
+                std::string ORF_file_path = tmp_dir + "colour_" + std::to_string(colour_ID) + "_ORFs.tmp";
                 std::ofstream ofs(ORF_file_path);
                 boost::archive::text_oarchive oa(ofs);
                 // write class instance to archive
@@ -622,7 +622,7 @@ std::pair<std::map<size_t, std::string>, std::map<size_t, std::string>> Graph::f
             {
                 const std::string base_filename = FM_fasta_file.substr(FM_fasta_file.find_last_of("/\\") + 1);
 
-                const auto idx_file_name = path_dir + "/" + base_filename + ".fmp";
+                const auto idx_file_name = path_dir + base_filename + ".fmp";
                 if (!load_from_file(fm_idx, idx_file_name))
                 {
                     cout << "FM-Index not available for " << FM_fasta_file << endl;
@@ -768,7 +768,7 @@ std::pair<std::map<size_t, std::string>, std::map<size_t, std::string>> Graph::f
             }
 
             {
-                std::string file_path = tmp_dir + "/colour_" + std::to_string(colour_ID) + "_edges.tmp";
+                std::string file_path = tmp_dir + "colour_" + std::to_string(colour_ID) + "_edges.tmp";
                 std::ofstream ofs(file_path);
                 boost::archive::text_oarchive oa(ofs);
                 // write class instance to archive
@@ -824,7 +824,7 @@ std::pair<RefindMap, bool> Graph::refind_gene(const size_t& colour_ID,
     {
         const std::string base_filename = FM_fasta_file.substr(FM_fasta_file.find_last_of("/\\") + 1);
 
-        const auto idx_file_name = path_dir + "/" + base_filename + ".fmp";
+        const auto idx_file_name = path_dir + base_filename + ".fmp";
         if (!load_from_file(fm_idx, idx_file_name))
         {
             cout << "FM-Index not available for " << FM_fasta_file << endl;

src/indexing.cpp

@@ -402,7 +402,7 @@ void calculate_genome_paths(const std::vector<Kmer>& head_kmer_arr,
     // create fm index file name
     const std::string base_filename = fasta_file.substr(fasta_file.find_last_of("/\\") + 1);
 
-    const auto idx_file_name = path_dir + "/" + base_filename + ".fmp";
+    const auto idx_file_name = path_dir + base_filename + ".fmp";
 
     // initialise string of nodes for FM-index generation
     std::string genome_path;

src/match_string.cpp

@@ -19,7 +19,7 @@ std::pair<fm_index_coll, std::vector<size_t>> index_fasta(const std::string& fas
     // create fm index file name
     const std::string base_filename = fasta_file.substr(fasta_file.find_last_of("/\\") + 1);
 
-    std::string idx_file_name = path_dir + "/" + base_filename + ".fms";
+    std::string idx_file_name = path_dir + base_filename + ".fms";
 
     // create entry for start and end of contigs within fm_index
     std::vector<size_t> contig_locs;