Merge pull request #219 from LadnerLab/deconv-ISS218

Deconv iss218

include/modules/deconv/module_deconv.h

@@ -1,12 +1,13 @@
 #ifndef MODULE_DECONV_HH_INCLUDED 
 #define MODULE_DECONV_HH_INCLUDED 
-#include <string>
-#include <vector>
 #include <algorithm>
-#include <set>
+#include <boost/algorithm/string.hpp>
+#include <map>
 #include <memory>
+#include <set>
+#include <string>
+#include <vector>
 #include <unordered_map>
-#include <map>
 #include <unordered_set>
 
 #include "overlap_data.h"
@@ -107,25 +108,36 @@ class module_deconv : public module
      * @note id is only used for summation scoring
      * @returns The score the id contributes to the species, as defined by 
      *          score_strat
-     **/
-    double score_peptide_for_species( const peptide& peptide,
-                                      std::unordered_map
-                                      <std::string,
-                                      std::vector<std::pair<std::string,double>
-                                      >>&
-                                      spec_count_map,
-                                      std::string id,
-                                      evaluation_strategy::score_strategy score_strat
-                                    );
-    /**
-     * Parse a map that will provide name->tax id mappings. This map should be formatted 
-     * in the same manner as that of 'lineage.dmp' from NCBI. 
-     * @param fname The name fo the file to parse
-     * @param name_map the destination map that will store the mappings of id->name
-     **/
-    void
-        parse_name_map( std::string fname, std::map<std::string,std::string>& name_map );
-
+     */
+    double score_peptide_for_species(
+        const peptide& peptide,
+        std::unordered_map<
+            std::string, std::vector<std::pair<std::string,double>>
+        >& spec_count_map,
+        std::string id,
+        evaluation_strategy::score_strategy score_strat
+    );
+
+    /**
+     * Parse a map which provides name->taxID mappings. This map should be
+     * formatted in the same manner as that of 'lineage.dmp' from NCBI. 
+     * @param fname Name of the file to parse
+     * @param name_map Destination map which stores mappings of ID->name
+     */
+    void parse_ncbi_name_map(
+        std::string fname, std::map<std::string, std::string>& name_map
+    );
+
+    /**
+     * Parse a map which providing name->tax ID mappings. This map shoudl be
+     * tab-delimited.
+     * @param fname Name of file to be parsed
+     * @param name_map Destination map which stores mappings of ID->name
+     */
+    void parse_custom_name_map(
+        std::tuple<std::string, std::string, std::string>& tup,
+        std::map<std::string, std::string>& name_map
+    );
 
     /**
      * Write output to a file that will be named 'out_name'
@@ -137,7 +149,7 @@ class module_deconv : public module
      *        used for outputting what the original count/score of each enriched species is.
      *        This is useful for seeing how the scores and counts have changed for each 
      *        of the enriched species.
-     **/
+     */
     void write_outputs( std::string out_name,
                         std::map<std::string,std::string>*
                         id_name_map,
@@ -146,7 +158,8 @@ class module_deconv : public module
                         >&
                         out_counts,
                         std::unordered_map<std::string,std::pair<double,double>>&
-                        original_scores
+                        original_scores,
+                        std::string custom_id_header
                       );
 
     /**
@@ -895,7 +908,8 @@ class module_deconv : public module
      * @param filename filepath to tab-delimited file
      * @returns map of taxID and threshold
      **/
-    void thresh_file_to_map( std::unordered_map<std::string, std::size_t>& thresh_map, std::string filename );
+    void thresh_file_to_map( std::unordered_map<std::string, std::size_t>& thresh_map, std::string filename, 
+                        const std::vector<std::pair<std::string, std::vector<std::pair<std::string, double>>>> pep_species_vec );
 };
 
 #endif // MODULE_DECONV_HH_INCLUDED 

include/modules/deconv/options_deconv.h

@@ -1,10 +1,13 @@
 #ifndef OPTIONS_DECONV_HH_INCLUDED
 #define OPTIONS_DECONV_HH_INCLUDED
 
+#include "logger.h"
 #include "options.h"
 
 #include <string>
+#include <tuple>
 #include <vector>
+#include <boost/algorithm/string.hpp>
 
 class options_deconv : public options
 {
@@ -120,9 +123,14 @@ class options_deconv : public options
     std::size_t k;
 
     /**
-     * The name of the file to
-     * write the id name map to.
-     **/
+     * Tuple containing information for parsing user-defined ID name map
+     */
+    std::tuple<std::string, std::string, std::string>
+        custom_id_name_map_info; //!< 0 = ID name map, 1 = Taxon ID column name, 2 = Sequence column name
+
+    /**
+     * Name of file containing NCBI taxID to taxon name mappings
+     */
     std::string id_name_map_fname;
 
     /**
@@ -131,6 +139,29 @@ class options_deconv : public options
      **/
     std::string enriched_file_ending;
 
+    /**
+     * Translates information from string 'info' into a tuple of length 3 and
+     * any type.
+     * Note: is probably only going to be used to move custom ID name map
+     * information.
+     * @param tup Reference to tuple which will be filled
+     * @param info Comma-delimited string of information
+     * @param opt_name Name of option for error output
+     */
+    // TODO: make into a generic function in options class,
+    // or a virtual method - whichever is more practical for the future
+    void set_info(
+        std::tuple<std::string, std::string, std::string>& tup,
+        std::string info,
+        std::string opt_name
+    );
+
+    /**
+     */
+    // TODO: make into a generic function in options class,
+    // or a virtual method - whichever is more practical for the future
+    std::string tuple_to_string(std::tuple<std::string, std::string, std::string>& tup);
 };
 
-#endif // OPTIONS_DECONV_HH_INCLUDED
+#endif /* OPTIONS_DECONV_HH_INCLUDED */
+