Merge pull request #84 from gordonrix/add-batch-mode-indexing

Add batch mode indexing to consensus sequence headers

Author: yangao07

src/abpoa.c

@@ -150,9 +150,12 @@ int abpoa_main(char *file_fn, int is_list, abpoa_para_t *abpt){
     abpoa_t *ab = abpoa_init();
     if (is_list) { // input file list
         FILE *list_fp = fopen(file_fn, "r"); char read_fn[1024];
+        int batch_index = 1;
         while (fgets(read_fn, sizeof(read_fn), list_fp)) {
             read_fn[strlen(read_fn)-1] = '\0';
+            abpt->batch_index = batch_index;
             abpoa_msa1(ab, abpt, read_fn, stdout);
+            batch_index++;
         }
         fclose(list_fp);
     } else // input file

src/abpoa.h

@@ -83,6 +83,7 @@ typedef struct {
     int align_mode, gap_mode, max_n_cons, cons_algrm; // consensus calling algorithm: 0: partial order graph, 1: majority voting
     double min_freq; // for multiploid data
     int verbose; // to control output msg
+    int batch_index; // index of current file in batch mode (for output header)
 
     // char LogTable65536[65536];
     // char bit_table16[65536];

src/abpoa_align.c

@@ -149,6 +149,7 @@ abpoa_para_t *abpoa_init_para(void) {
     abpt->progressive_poa = 0; // progressive partial order alignment
 
     abpt->verbose = ABPOA_NONE_VERBOSE;
+    abpt->batch_index = 0; // 0 means not in batch mode
 
     // abpt->simd_flag = simd_check();
 

src/abpoa_output.c

@@ -579,6 +579,9 @@ void abpoa_output_fx_consensus(abpoa_t *ab, abpoa_para_t *abpt, FILE *out_fp) {
     for (cons_i = 0; cons_i < abc->n_cons; ++cons_i) {
         if (abpt->out_fq) fprintf(out_fp, "@Consensus_sequence");
         else fprintf(out_fp, ">Consensus_sequence");
+        if (abpt->batch_index > 0) {
+            fprintf(out_fp, "_%d", abpt->batch_index); // batch index for file mapping
+        }
         if (abc->n_cons > 1) {
             fprintf(out_fp, "_%d ", cons_i+1); // cons_id
             for (j = 0; j < abc->clu_n_seq[cons_i]; ++j) { // cluter read_id
@@ -592,6 +595,9 @@ void abpoa_output_fx_consensus(abpoa_t *ab, abpoa_para_t *abpt, FILE *out_fp) {
         } fprintf(out_fp, "\n");
         if (abpt->out_fq) {
             fprintf(out_fp, "+Consensus_sequence");
+            if (abpt->batch_index > 0) {
+                fprintf(out_fp, "_%d", abpt->batch_index); // batch index for file mapping
+            }
             if (abc->n_cons > 1) {
                 fprintf(out_fp, "_%d ", cons_i+1); // cons_id
                 for (j = 0; j < abc->clu_n_seq[cons_i]; ++j) { // cluter read_id