Added extension mode to align_benchmark (for testing)

smarco · Sep 25, 2023 · d09e0b2 · d09e0b2
1 parent 9fdf46e
commit d09e0b2
Show file tree

Hide file tree

Showing 13 changed files with 77 additions and 60 deletions.
diff --git a/alignment/cigar.h b/alignment/cigar.h
@@ -32,6 +32,7 @@
 #ifndef CIGAR_H_
 #define CIGAR_H_
 
+#include <stdbool.h>
 #include "system/mm_allocator.h"
 #include "alignment/linear_penalties.h"
 #include "alignment/affine_penalties.h"

diff --git a/system/mm_allocator.h b/system/mm_allocator.h
@@ -35,6 +35,7 @@
 #ifndef MM_ALLOCATOR_H_
 #define MM_ALLOCATOR_H_
 
+#include <stdbool.h>
 #include "utils/vector.h"
 
 /*

diff --git a/tests/wfa.utest.performance.sh b/tests/wfa.utest.performance.sh
@@ -38,4 +38,7 @@ rm $OUTPUT/*.log $OUTPUT/*.alg
 \time -v ./bin/align_benchmark -a $ALGORITHM -i ../data/sim.l100K.n1.e10.seq -o $OUTPUT/sim.l100K.e10.Wb.alg $BIWFA      &> $OUTPUT/sim.l100K.e10.Wb.log
 
 # Run the check
-./tests/wfa.utest.cmp.sh tests/ tests/wfa.utest.performance.check/ --cmp-performance
+echo ">>> "
+echo ">>> Performance [base vs new]: "
+echo ">>> "
+./tests/wfa.utest.cmp.sh tests/wfa.utest.performance.check/ tests/ --cmp-performance
diff --git a/tests/wfa.utest.sh b/tests/wfa.utest.sh
@@ -90,9 +90,9 @@ grep "Time.Alignment" $LOG | awk '{if ($4 != "ms") print $3" "$4}' | sort -n > $
 grep "Maximum resident set size" $LOG | awk '{print $6}' | sort -n > $LOG.mem
 
 # Display performance
-echo ">>> Performance Time (s): "
+echo ">>> Performance Time (s) [base vs new]: "
 paste <(tail -n 4 $OUTPUT/wfa.utest.check/wfa.utest.log.time) <(tail -n 4 $OUTPUT/wfa.utest.log.time) 
-echo ">>> Performance Mem (KB): "
+echo ">>> Performance Mem (KB) [base vs new]: "
 paste <(tail -n 4 $OUTPUT/wfa.utest.check/wfa.utest.log.mem) <(tail -n 4 $OUTPUT/wfa.utest.log.mem) 
 
 # Display correct

diff --git a/tools/align_benchmark/align_benchmark.c b/tools/align_benchmark/align_benchmark.c
@@ -217,7 +217,16 @@ wavefront_aligner_t* align_input_configure_wavefront(
       break;
   }
   // Select alignment form
-  attributes.alignment_form.span = (parameters.endsfree) ? alignment_endsfree : alignment_end2end;
+  if (parameters.align_span_extension) {
+    attributes.alignment_form.span = alignment_endsfree;
+    attributes.alignment_form.extension = true;
+  } else if (parameters.align_span_endsfree) {
+    attributes.alignment_form.span = alignment_endsfree;
+    attributes.alignment_form.extension = false;
+  } else { // Global
+    attributes.alignment_form.span = alignment_end2end;
+    attributes.alignment_form.extension = false;
+  }
   // Misc
   attributes.plot.enabled = (parameters.plot != 0);
   attributes.plot.align_level = (parameters.plot < 0) ? -1 : parameters.plot - 1;
@@ -236,8 +245,6 @@ void align_input_configure_global(
   align_input->linear_penalties = parameters.linear_penalties;
   align_input->affine_penalties = parameters.affine_penalties;
   align_input->affine2p_penalties = parameters.affine2p_penalties;
-  // Alignment form
-  align_input->ends_free = parameters.endsfree;
   // Output
   align_input->output_file = parameters.output_file;
   align_input->output_full = parameters.output_full;
@@ -271,7 +278,7 @@ void align_input_configure_global(
 void align_input_configure_local(
     align_input_t* const align_input) {
   // Ends-free configuration
-  if (parameters.endsfree) {
+  if (parameters.align_span_endsfree) {
     const int plen = align_input->pattern_length;
     const int tlen = align_input->text_length;
     align_input->pattern_begin_free = nominal_prop_u32(plen,parameters.pattern_begin_free);

diff --git a/tools/align_benchmark/align_benchmark_params.c b/tools/align_benchmark/align_benchmark_params.c
@@ -68,7 +68,9 @@ align_bench_params_t parameters = {
       .gap_extension2 = 1,
   },
   // Alignment form
-  .endsfree = false,
+  .align_span_global = true,
+  .align_span_extension = false,
+  .align_span_endsfree = false,
   .pattern_begin_free = 0.0,
   .text_begin_free = 0.0,
   .pattern_end_free = 0.0,
@@ -134,10 +136,10 @@ void usage() {
       "          --linear-penalties|p M,X,I                                    \n"
       "          --affine-penalties|g M,X,O,E                                  \n"
       "          --affine2p-penalties M,X,O1,E1,O2,E2                          \n"
-      "          --ends-free P0,Pf,T0,Tf                                       \n"
       "        [Wavefront parameters]                                          \n"
       "          --wfa-score-only                                              \n"
-      "          --wfa-memory-mode 'high'|'med'|'low'|'ultralow'               \n"
+      "          --wfa-span 'global'|'extension'|'ends-free[,P0,Pf,T0,Tf]'     \n"
+      "          --wfa-memory 'high'|'med'|'low'|'ultralow'                    \n"
       "          --wfa-heuristic STRATEGY                                      \n"
       "          --wfa-heuristic-parameters  P1,P2[,P3]                        \n"
       "            [STRATEGY='banded-static']                                  \n"
@@ -191,12 +193,12 @@ void parse_arguments(
     { "linear-penalties", required_argument, 0, 'p' },
     { "affine-penalties", required_argument, 0, 'g' },
     { "affine2p-penalties", required_argument, 0, 900 },
-    { "ends-free", required_argument, 0, 901 },
     /* Wavefront parameters */
     { "wfa-score-only", no_argument, 0, 1000 },
-    { "wfa-memory-mode", required_argument, 0, 1001 },
-    { "wfa-heuristic", required_argument, 0, 1002 },
-    { "wfa-heuristic-parameters", required_argument, 0, 1003 },
+    { "wfa-span", required_argument, 0, 1001 },
+    { "wfa-memory", required_argument, 0, 1002 },
+    { "wfa-heuristic", required_argument, 0, 1003 },
+    { "wfa-heuristic-parameters", required_argument, 0, 1004 },
     { "wfa-max-memory", required_argument, 0, 1005 },
     { "wfa-max-steps", required_argument, 0, 1006 },
     { "wfa-max-threads", required_argument, 0, 1007 },
@@ -322,25 +324,44 @@ void parse_arguments(
       parameters.affine2p_penalties.gap_extension2 = atoi(sentinel);
       break;
     }
-    case 901: { // --ends-free P0,Pf,T0,Tf
-      parameters.endsfree = true;
-      char* sentinel = strtok(optarg,",");
-      parameters.pattern_begin_free = atof(sentinel);
-      sentinel = strtok(NULL,",");
-      parameters.pattern_end_free = atof(sentinel);
-      sentinel = strtok(NULL,",");
-      parameters.text_begin_free = atof(sentinel);
-      sentinel = strtok(NULL,",");
-      parameters.text_end_free = atof(sentinel);
-      break;
-    }
     /*
      * Wavefront parameters
      */
     case 1000: // --wfa-score-only
       parameters.wfa_score_only = true;
       break;
-    case 1001: // --wfa-memory-mode in {'high','med','low'}
+    case 1001: { // --wfa-span P0,Pf,T0,Tf
+      if (strcmp(optarg,"global")==0) {
+        parameters.align_span_global = true;
+        parameters.align_span_extension = false;
+        parameters.align_span_endsfree = false;
+      } else if (strcmp(optarg,"extension")==0 ||
+                 strcmp(optarg,"extend")==0) {
+        parameters.align_span_global = false;
+        parameters.align_span_extension = true;
+        parameters.align_span_endsfree = false;
+      } else if (strcmp(optarg,"ends-free")==0 ||
+                 strcmp(optarg,"endsfree")==0) {
+        parameters.align_span_global = false;
+        parameters.align_span_extension = false;
+        parameters.align_span_endsfree = true;
+        // Parse arguments
+        char* sentinel = strtok(optarg,",");
+        sentinel = strtok(NULL,","); // Skip span-keyword
+        parameters.pattern_begin_free = atof(sentinel);
+        sentinel = strtok(NULL,",");
+        parameters.pattern_end_free = atof(sentinel);
+        sentinel = strtok(NULL,",");
+        parameters.text_begin_free = atof(sentinel);
+        sentinel = strtok(NULL,",");
+        parameters.text_end_free = atof(sentinel);
+      } else {
+        fprintf(stderr,"Option '--wfa-span' must be in {'global','extension','ends-free'}\n");
+        exit(1);
+      }
+      break;
+    }
+    case 1002: // --wfa-memory in {'high','med','low'}
       if (strcmp(optarg,"high")==0) {
         parameters.wfa_memory_mode = wavefront_memory_high;
       } else if (strcmp(optarg,"med")==0) {
@@ -350,11 +371,11 @@ void parse_arguments(
       } else if (strcmp(optarg,"ultralow")==0) {
         parameters.wfa_memory_mode = wavefront_memory_ultralow;
       } else {
-        fprintf(stderr,"Option '--wfa-memory-mode' must be in {'high','med','low','ultralow'}\n");
+        fprintf(stderr,"Option '--wfa-memory' must be in {'high','med','low','ultralow'}\n");
         exit(1);
       }
       break;
-    case 1002: // --wfa-heuristic in {'none'|'banded-static'|'banded-adaptive'|'wfa-adaptive'|'xdrop'|'zdrop'}
+    case 1003: // --wfa-heuristic in {'none'|'banded-static'|'banded-adaptive'|'wfa-adaptive'|'xdrop'|'zdrop'}
       if (strcmp(optarg,"none")==0) {
         parameters.wfa_heuristic = wf_heuristic_none;
       } else if (strcmp(optarg,"banded-static")==0 || strcmp(optarg,"banded")==0) {
@@ -372,7 +393,7 @@ void parse_arguments(
         exit(1);
       }
       break;
-    case 1003: { // --wfa-heuristic-parameters  <P1>,<P2>[,<P3>]
+    case 1004: { // --wfa-heuristic-parameters  <P1>,<P2>[,<P3>]
       char* sentinel = strtok(optarg,",");
       const int p1 = atoi(sentinel);
       parameters.wfa_heuristic_p1 = p1;
@@ -493,19 +514,16 @@ void parse_arguments(
     exit(1);
   }
   // Check 'ends-free' parameter
-  if (parameters.endsfree) {
+  if (parameters.align_span_endsfree || parameters.align_span_extension) {
     switch (parameters.algorithm) {
-      case alignment_gap_affine_swg:
-        parameters.algorithm = alignment_gap_affine_swg_endsfree;
-        break;
       case alignment_indel_wavefront:
       case alignment_edit_wavefront:
       case alignment_gap_linear_wavefront:
       case alignment_gap_affine_wavefront:
       case alignment_gap_affine2p_wavefront:
         break;
       default:
-        fprintf(stderr,"Ends-free variant not implemented for the selected algorithm\n");
+        fprintf(stderr,"Ends-free/Extension variant not implemented for the selected algorithm\n");
         exit(1);
         break;
     }

diff --git a/tools/align_benchmark/align_benchmark_params.h b/tools/align_benchmark/align_benchmark_params.h
@@ -81,8 +81,10 @@ typedef struct {
   linear_penalties_t linear_penalties;
   affine_penalties_t affine_penalties;
   affine2p_penalties_t affine2p_penalties;
-  // Alignment form
-  bool endsfree;
+  // Alignment span
+  bool align_span_global;
+  bool align_span_extension;
+  bool align_span_endsfree;
   double pattern_begin_free;
   double text_begin_free;
   double pattern_end_free;

diff --git a/tools/align_benchmark/benchmark/benchmark_utils.c b/tools/align_benchmark/benchmark/benchmark_utils.c
@@ -42,7 +42,6 @@
 void benchmark_align_input_clear(
     align_input_t* const align_input) {
   // Alignment form
-  align_input->ends_free = false;
   align_input->pattern_begin_free = 0;
   align_input->text_begin_free = 0;
   align_input->pattern_end_free = 0;

diff --git a/tools/align_benchmark/benchmark/benchmark_utils.h b/tools/align_benchmark/benchmark/benchmark_utils.h
@@ -69,7 +69,6 @@ typedef struct {
   affine_penalties_t affine_penalties;
   affine2p_penalties_t affine2p_penalties;
   // Alignment form
-  bool ends_free;
   int pattern_begin_free;
   int text_begin_free;
   int pattern_end_free;

diff --git a/utils/vector.h b/utils/vector.h
@@ -34,6 +34,7 @@
 #define VECTOR_H_
 
 #include <stdint.h>
+#include <stdbool.h>
 
 /*
  * Checkers

diff --git a/wavefront/wavefront_debug.c b/wavefront/wavefront_debug.c
@@ -133,6 +133,7 @@ void wavefront_report_lite(
   const int text_length = sequences->text_length;
   const int status = wf_aligner->align_status.status;
   const uint64_t memory_used = wf_aligner->align_status.memory_used;
+  const bool has_cigar = alignment_completed && !cigar_is_null(wf_aligner->cigar);
   // BANNER (#0)
   if (alignment_completed) {
     fprintf(stream,"[WFA::Debug]");
@@ -145,6 +146,10 @@ void wavefront_report_lite(
   const int score = wf_aligner->cigar->score;
   if (alignment_completed && score!=INT32_MIN) {
     fprintf(stream,"\t%d",score);
+    if (has_cigar) {
+      const int edit_dist = cigar_score_edit(wf_aligner->cigar);
+      fprintf(stream,"/%1.2f",(float)edit_dist/(float)MAX(pattern_length,text_length));
+    }
   } else {
     fprintf(stream,"\t*");
   }
@@ -184,7 +189,7 @@ void wavefront_report_lite(
       wf_aligner->wf_components.historic_max_hi);
   fprintf(stream,"]\t");
   // CIGAR (#8)
-  if (!alignment_completed || cigar_is_null(wf_aligner->cigar)) {
+  if (!has_cigar) {
     fprintf(stream,"-");
   } else {
     cigar_print(stream,wf_aligner->cigar,true);

diff --git a/wavefront/wavefront_penalties.h b/wavefront/wavefront_penalties.h
@@ -89,25 +89,6 @@ void wavefront_penalties_set_affine2p(
     wavefront_penalties_t* const wf_penalties,
     affine2p_penalties_t* const affine2p_penalties);
 
-/*
- * Score conversion
- */
-int wavefront_penalties_get_score_indel(
-    wavefront_penalties_t* const wf_penalties,
-    const int score);
-int wavefront_penalties_get_score_edit(
-    wavefront_penalties_t* const wf_penalties,
-    const int score);
-int wavefront_penalties_get_score_linear(
-    wavefront_penalties_t* const wf_penalties,
-    const int score);
-int wavefront_penalties_get_score_affine(
-    wavefront_penalties_t* const wf_penalties,
-    const int score);
-int wavefront_penalties_get_score_affine2p(
-    wavefront_penalties_t* const wf_penalties,
-    const int score);
-
 /*
  * Display
  */

diff --git a/wavefront/wavefront_sequences.c b/wavefront/wavefront_sequences.c
@@ -73,7 +73,7 @@ void wavefront_sequences_init_allocate(
     if (wf_sequences->seq_buffer != NULL) free(wf_sequences->seq_buffer);
     // Allocate
     const int proposed_size = buffer_size + buffer_size/2;
-    wf_sequences->seq_buffer = calloc(proposed_size, 1);
+    wf_sequences->seq_buffer = calloc(proposed_size,1);
     wf_sequences->seq_buffer_allocated = proposed_size;
   }
   // Assign memory