Fixes segmentation fault when building SA index (#11), and improves t…

…he help display for build index commands

src/build-index/index_builder.c

@@ -1,6 +1,5 @@
 #include "index_builder.h"
 
-
 //------------------------------------------------------------------------------------
 //------------------------------------------------------------------------------------
 
@@ -36,9 +35,11 @@ void index_options_free(index_options_t *options) {
 //------------------------------------------------------------------------------------
 
 void** argtable_index_options_new(int mode) {
-  int num_options = NUM_INDEX_OPTIONS;
+  int num_options;
   if (mode == BWT_INDEX) { 
-    num_options += NUM_INDEX_BWT_OPTIONS; 
+    num_options = NUM_INDEX_BWT_OPTIONS; 
+  } else {
+    num_options = NUM_INDEX_SA_OPTIONS; 
   }
 
   // NUM_OPTIONS +1 to allocate end structure
@@ -98,29 +99,51 @@ index_options_t *read_CLI_index_options(void **argtable, index_options_t *option
 //------------------------------------------------------------------------------------
 
 void usage_index(void **argtable, int mode) {
-  printf("\nUsage:\n\t%s %s <options>\n", HPG_ALIGNER_BIN, 
+  printf("\n");
+  printf("+===============================================================+\n");
+  if (mode == BWT_INDEX) {
+    printf("|                HPG-Aligner help for building BWT index        |\n");
+  } else {
+    printf("|                HPG-Aligner help for building SA index         |\n");
+  }
+  printf("+===============================================================+\n");
+  printf("Usage:\n");
+  printf("\t%s %s -g|--ref-genome=<file> -i|--index=<file> [options]\n", 
+	 HPG_ALIGNER_BIN,
 	 (mode == BWT_INDEX ? "build-bwt-index" : "build-sa-index"));
 
-  //arg_print_syntaxv(stdout, argtable, "\n");
-  printf("\nOptions:\n");
-  arg_print_glossary(stdout, argtable, "\t%-50s\t%s\n");
-
-  exit(0);
+  printf("\n");
+  printf("Mandatory parameters:\n");
+  printf("\t-g, --ref-genome              Reference genome (FASTA format)\n");
+  printf("\t-i, --index=<file>            Index directory name\n");
+  printf("\n");
+  printf("Options:\n");
+  printf("\t-a, --alternative-map=<file>  Alternative mapping filename. This two-columns file contains the alternative sequence names with their corresponding chromosome names (only for SA index)\n");
+  printf("\t-d, --decoy-genome=<file>     Decoy genome in FASTA format (only for SA index)\n");
+  printf("\t-v, --version                 Display version\n");
+  printf("\t-h, --help                    Help option\n");
 }
 
 //------------------------------------------------------------------------------------
 
 index_options_t *parse_index_options(int argc, char **argv) {
-  int mode = SA_INDEX, num_options = NUM_INDEX_OPTIONS;
+  int mode, num_options;
 
   if (strcmp(argv[0], "build-bwt-index") == 0) {
     mode = BWT_INDEX;
-    num_options += NUM_INDEX_BWT_OPTIONS;
+    num_options = NUM_INDEX_BWT_OPTIONS;
   } else if (strcmp(argv[0], "build-sa-index") == 0) {
     mode = SA_INDEX;
-  } 
+    num_options = NUM_INDEX_SA_OPTIONS;
+  } else {
+    fprintf(stdout, "\nErrors:\n");
+    printf("\tUnknown command: %s\n", argv[0]);
+    usage_index(NULL, mode);
+    exit(-1);
+  }
 
   void **argtable = argtable_index_options_new(mode);
+
   index_options_t *options = index_options_new();
   if (argc < 2) {
     usage_index(argtable, mode);

src/build-index/index_builder.h

@@ -15,8 +15,8 @@
 #define SA_INDEX  0
 #define BWT_INDEX 1
 
-#define NUM_INDEX_OPTIONS     6
-#define NUM_INDEX_BWT_OPTIONS 0
+#define NUM_INDEX_SA_OPTIONS  6
+#define NUM_INDEX_BWT_OPTIONS 5
 
 #define BWT_RATIO_DEFAULT  8
 

src/sa/sa_index3.c

@@ -97,12 +97,12 @@ char *alt_names_display(alt_names_t *p) {
 //--------------------------------------------------------------------------------------
 
 sa_genome3_t *read_genome3(char *genome_filename) {
-  return read_genome3_ex(genome_filename, NULL);
+  return read_genome3_alt(genome_filename, NULL);
 }
 
 //--------------------------------------------------------------------------------------
 
-sa_genome3_t *read_genome3_ex(char *genome_filename, char *alt_filename) {
+sa_genome3_t *read_genome3_alt(char *genome_filename, char *alt_filename) {
 
   const int MAX_CHROM_NAME_LENGHT = 1024;
   uint reading_name, seq_name_count = 0;
@@ -289,6 +289,7 @@ sa_genome3_t *read_genome3_ex(char *genome_filename, char *alt_filename) {
   // update chromosomes for ALT sequences: chromosomes and flanks
   char *chrom_name;
   for (size_t i = 0; i < num_seqs; i++) {
+    seq_chroms[i] = i;
     if (seq_flags[i] == ALT_FLAG) {
       chrom_name = alt_names_get_chrom_name(seq_names[i], alt_names);
       if (chrom_name) {
@@ -1201,7 +1202,7 @@ void sa_index3_build_k18_alt(char *genome_filename, char *alt_filename,
   sprintf(filename_tab, "%s/%s.S", sa_index_dirname, prefix);
   printf("\nreading file genome %s...\n", genome_filename);
   gettimeofday(&start, NULL);
-  sa_genome3_t *genome = read_genome3_ex(genome_filename, alt_filename);
+  sa_genome3_t *genome = read_genome3_alt(genome_filename, alt_filename);
   gettimeofday(&stop, NULL);
 
   if (genome->length > MAX_GENOME_LENGTH || genome->num_seqs > MAX_NUM_SEQUENCES) {
@@ -1215,7 +1216,7 @@ void sa_index3_build_k18_alt(char *genome_filename, char *alt_filename,
     exit(-1);
   }
 
-  //sa_genome3_display(genome);
+  //  sa_genome3_display(genome);
 
   // write S to file
   f_tab = fopen(filename_tab, "wb");
@@ -1525,7 +1526,7 @@ void sa_index3_build_k18_alt(char *genome_filename, char *alt_filename,
   fprintf(f_tab, "6. Number of suffixes\n");
   fprintf(f_tab, "7. Genome length\n");
   fprintf(f_tab, "8. Number of sequencess\n");
-  fprintf(f_tab, "9. One line per sequence: name, length, type, chrom, start, end, left and right flanks (the last five fields for HAP sequences)\n");
+  fprintf(f_tab, "9. One line per sequence: name, length, type, chrom, start, end, left and right flanks (the last five fields for ALT sequences)\n");
   fclose(f_tab);
 
   sprintf(filename_tab, "%s/index", sa_index_dirname);

src/sa/sa_index3.h

@@ -69,7 +69,7 @@ typedef struct sa_genome3 {
 //--------------------------------------------------------------------------------------
 
 sa_genome3_t *read_genome3(char *genome_filename);
-sa_genome3_t *read_genome3_ex(char *genome_filename, char *alt_filename);
+sa_genome3_t *read_genome3_alt(char *genome_filename, char *alt_filename);
 
 //--------------------------------------------------------------------------------------
 
@@ -83,6 +83,9 @@ static inline sa_genome3_t *sa_genome3_new(size_t length, size_t num_seqs,
   p->num_seqs = num_seqs;
   p->seq_lengths = seq_lengths;
   p->seq_flags = seq_flags;
+  p->seq_chroms = seq_chroms;
+  p->seq_starts = seq_starts;
+  p->seq_ends = seq_ends;
   if (num_seqs && seq_lengths) {
     p->seq_offsets = (size_t *) calloc(num_seqs, sizeof(size_t));
     size_t offset = 0;
@@ -105,6 +108,8 @@ static inline sa_genome3_t *sa_genome3_new(size_t length, size_t num_seqs,
   size_t flank_size;
   char *alt_seq, *chrom_seq;
   for (size_t i = 0; i < num_seqs; i++) {
+    left_flanks[i] = 0;
+    right_flanks[i] = 0;
     if (seq_flags[i] == ALT_FLAG) {
       // calculate left flank
       flank_size = 0;
@@ -195,7 +200,7 @@ static inline void sa_genome3_display(sa_genome3_t *p) {
     printf("%u\t%s\t%s\t%lu\t%lu\t%lu\t%s\t%lu\t%lu\t%lu\t%lu\n", 
 	   i, GET_SEQ_FLAG_NAME(p->seq_flags[i]), p->seq_names[i],
 	   p->seq_lengths[i], p->seq_offsets[i],
-	   p->seq_chroms[i], (p->seq_flags[i] == ALT_FLAG ? p->seq_names[p->seq_chroms[i]] : ""), 
+	   p->seq_chroms[i], p->seq_names[p->seq_chroms[i]], 
 	   p->seq_starts[i], p->seq_ends[i],
 	   p->left_flanks[i], p->right_flanks[i]);
   }