Merge pull request #29 from veg/develop

Adding delimiter options for tn93

CMakeLists.txt

@@ -12,8 +12,8 @@ set(CMAKE_CONFIGURATION_TYPES Release)
 # SET VERSION FROM FILE
 file (STRINGS "VERSION.txt" VERSION_NUMBER)
 add_definitions (-DVERSION_NUMBER=\"${VERSION_NUMBER}\")
-add_compile_options (-O3 -std=c++14 -march=native -funroll-loops )
-add_link_options(-O3 -std=c++14 -march=native -funroll-loops)
+add_compile_options (-O3 -std=c++14  -funroll-loops )
+add_link_options(-O3 -std=c++14 -funroll-loops)
 
 include_directories(
     src/

VERSION.txt

@@ -1 +1 @@
-v1.0.9
+v1.0.10

src/TN93.cpp

@@ -263,9 +263,9 @@ int main(int argc, const char *argv[]) {
   if (args.format != hyphy) {
     if (args.do_count == false) {
       if (args.format == csv)
-        fprintf(args.output, "ID1,ID2,Distance\n");
+        fprintf(args.output, "ID1%cID2%cDistance\n", args.delimiter, args.delimiter);
       else
-        fprintf(args.output, "Seq1,Seq2,Distance\n");
+        fprintf(args.output, "Seq1%cSeq2%cDistance\n", args.delimiter, args.delimiter);
     }
   } else {
     distanceMatrix = new double[sequenceCount * sequenceCount];
@@ -375,11 +375,11 @@ int main(int argc, const char *argv[]) {
           // to self as 0
           if (args.format == csv) {
 #pragma omp critical
-            fprintf(args.output, "%s,%s,%g\n", n1, n1, 0.0);
+            fprintf(args.output, "%s%c%s%c%g\n", n1, args.delimiter, n1, args.delimiter, 0.0);
           } else {
             if (args.format == csvn) {
 #pragma omp critical
-              fprintf(args.output, "%ld,%ld,%g\n", mapped_id, mapped_id, 0.0);
+              fprintf(args.output, "%ld%c%ld%c%g\n", mapped_id, args.delimiter, mapped_id, args.delimiter, 0.0);
             }
           }
         }
@@ -425,12 +425,12 @@ int main(int argc, const char *argv[]) {
           if (!args.do_count) {
             if (args.format == csv) {
 #pragma omp critical
-              fprintf(args.output, "%s,%s,%g\n", n1,
-                      stringText(names, nameLengths, mapped_id2), thisD);
+              fprintf(args.output, "%s%c%s%c%g\n", n1,args.delimiter,
+                      stringText(names, nameLengths, mapped_id2), args.delimiter,thisD);
             } else {
               if (args.format == csvn) {
 #pragma omp critical
-                fprintf(args.output, "%ld,%ld,%g\n", mapped_id, mapped_id2,
+                fprintf(args.output, "%ld%c%ld%c%g\n", mapped_id, args.delimiter, mapped_id2, args.delimiter,
                         thisD);
 
               } else {

src/argparse.cpp

@@ -31,6 +31,7 @@ namespace argparse
   "[-c] "
   "[-0] "
   "[-q] "
+  "[-D DELIMITER]"
   "[FASTA]\n";
 
   const char help_msg[] =
@@ -68,6 +69,7 @@ namespace argparse
   "  -m                       compute inter- and intra-population means suitable for FST caclulations\n"
   "                           only applied when -s is used to provide a second file\n"
   "  -u PROBABILITY           subsample sequences with specified probability (a value between 0 and 1, default = " TO_STR ( DEFAULT_INCLUDE_PROB) ")\n"
+  "  -D DELIMITER             use this character as a delimiter in the output column-file (a character, default = " TO_STR ( DEFAULT_DELIMITER) ")\n"
   "  -0                       report distances between each sequence and itself (as 0); this is useful to ensure every sequence\n"
   "                           in the input file appears in the output, e.g. for network construction to contrast clustered/unclustered\n"
   "  -q                       do not report progress updates and other diagnostics to stderr \n"
@@ -126,7 +128,8 @@ namespace argparse
   counts_in_name ( DEFAULT_COUNTS_IN_NAME ),
   include_prob( DEFAULT_INCLUDE_PROB ),
   ambigs_to_resolve(NULL),
-  resolve_fraction(DEFAULT_FRACTION)
+  resolve_fraction(DEFAULT_FRACTION),
+  delimiter (DEFAULT_DELIMITER)
   {
       // skip arg[0], it's just the program name
     for (int i = 1; i < argc; ++i ) {
@@ -149,6 +152,7 @@ namespace argparse
         else if (  arg[1] == 's')  parse_second_in( next_arg (i, argc, argv) );
         else if (  arg[1] == 'd')  parse_counts_in_name( next_arg (i, argc, argv) );
         else if (  arg[1] == 'u')  parse_include_prob( next_arg (i, argc, argv) );
+        else if (  arg[1] == 'D')  parse_delimiter ( next_arg (i, argc, argv) );
         else if (  arg[1] == 'b')  parse_bootstrap();
         else if (  arg[1] == 'r')  parse_bootstrap_two_files ();
         else if (  arg[1] == 'c')  parse_count();
@@ -275,19 +279,29 @@ namespace argparse
       strcpy (ambigs_to_resolve, str);
     }
   }
+
+  void args_t::parse_delimiter( const char * str )
+    {
+      if (strlen (str) == 1) {
+          delimiter = str[0];
+      } else {
+          ERROR( "invalid output format: %s", str );
+      }
+    }
 
-  void args_t::parse_format( const char * str )
-  {
-    if (!strcmp (str, "csv")) {
-      format = csv;
-    } else if (!strcmp (str, "csvn")) {
-      format = csvn;
-    } else if (!strcmp (str, "hyphy")) {
-      format = hyphy;
-    } else  {
-      ERROR( "invalid output format: %s", str );
+
+    void args_t::parse_format( const char * str )
+    {
+      if (!strcmp (str, "csv")) {
+        format = csv;
+      } else if (!strcmp (str, "csvn")) {
+        format = csvn;
+      } else if (!strcmp (str, "hyphy")) {
+        format = hyphy;
+      } else  {
+        ERROR( "invalid output format: %s", str );
+      }
     }
-  }
 
   void args_t::parse_count()
   {

src/argparse.hpp

@@ -12,6 +12,7 @@
 #define DEFAULT_COUNTS_IN_NAME    ':'
 #define DEFAULT_OVERLAP           100
 #define DEFAULT_INCLUDE_PROB      1.0
+#define DEFAULT_DELIMITER         ','
 
 #ifndef VERSION_NUMBER
 #define VERSION_NUMBER            "UNKNOWN"
@@ -56,6 +57,7 @@ namespace argparse
         double          include_prob;
         char            *ambigs_to_resolve;
         double          resolve_fraction;
+        char            delimiter;
 
         args_t( int, const char ** );
         ~args_t();
@@ -69,6 +71,7 @@ namespace argparse
         void parse_overlap  ( const char * );
         void parse_format   ( const char * );
         void parse_ambig    ( const char * );
+        void parse_delimiter      ( const char * );
         void parse_include_prob   ( const char * );
         void parse_counts_in_name
                             ( const char * );