Stripped version: very fast

c-core/setup_viterbi.c

@@ -0,0 +1,103 @@
+#include "protein.h"
+#include "protein_node.h"
+#include "vit.h"
+#include <math.h>
+
+struct viterbi_xtrans
+{
+  float const SB;
+  float const SN;
+  float const NN;
+  float const NB;
+
+  float const ET;
+  float const EC;
+  float const CC;
+  float const CT;
+
+  float const EB;
+  float const EJ;
+  float const JJ;
+  float const JB;
+
+  float const ME;
+  float const DE;
+};
+
+struct viterbi_xtrans viterbi_xtrans_init(struct xtrans x)
+{
+  return (struct viterbi_xtrans){
+      .SB = x.NB,
+      .SN = x.NN,
+      .NN = x.NN,
+      .NB = x.NB,
+
+      .ET = x.EC + x.CT,
+      .EC = x.EC + x.CC,
+      .CC = x.CC,
+      .CT = x.CT,
+
+      .EB = x.EJ + x.JB,
+      .EJ = x.EJ + x.JJ,
+      .JJ = x.JJ,
+      .JB = x.JB,
+
+      .ME = 0.0f,
+      .DE = 0.0f,
+  };
+}
+
+int setup_viterbi(struct vit *x, struct protein const *protein)
+{
+  int K = protein->core_size;
+  int rc = vit_setup(x, K);
+  if (rc) return rc;
+
+  struct viterbi_xtrans const xt = viterbi_xtrans_init(protein->xtrans);
+  vit_set_extr_trans(x, EXTR_TRANS_RR, -protein->null.RR);
+  vit_set_extr_trans(x, EXTR_TRANS_SN, -xt.SN);
+  vit_set_extr_trans(x, EXTR_TRANS_NN, -xt.NN);
+  vit_set_extr_trans(x, EXTR_TRANS_SB, -xt.SB);
+  vit_set_extr_trans(x, EXTR_TRANS_NB, -xt.NB);
+  vit_set_extr_trans(x, EXTR_TRANS_EB, -xt.EB);
+  vit_set_extr_trans(x, EXTR_TRANS_JB, -xt.JB);
+  vit_set_extr_trans(x, EXTR_TRANS_EJ, -xt.EJ);
+  vit_set_extr_trans(x, EXTR_TRANS_JJ, -xt.JJ);
+  vit_set_extr_trans(x, EXTR_TRANS_EC, -xt.EC);
+  vit_set_extr_trans(x, EXTR_TRANS_CC, -xt.CC);
+  vit_set_extr_trans(x, EXTR_TRANS_ET, -xt.ET);
+  vit_set_extr_trans(x, EXTR_TRANS_CT, -xt.CT);
+
+  for (int k = 0; k < K; ++k)
+  {
+    vit_set_core_trans(x, CORE_TRANS_BM, -protein->BMk[k], k);
+  }
+
+  vit_set_core_trans(x, CORE_TRANS_MM, INFINITY, 0);
+  vit_set_core_trans(x, CORE_TRANS_MD, INFINITY, 0);
+  vit_set_core_trans(x, CORE_TRANS_IM, INFINITY, 0);
+  vit_set_core_trans(x, CORE_TRANS_DM, INFINITY, 0);
+  vit_set_core_trans(x, CORE_TRANS_DD, INFINITY, 0);
+  for (int k = 0; k < K - 1; ++k)
+  {
+    vit_set_core_trans(x, CORE_TRANS_MM, -protein->nodes[k].trans.MM, k + 1);
+    vit_set_core_trans(x, CORE_TRANS_MI, -protein->nodes[k].trans.MI, k + 0);
+    vit_set_core_trans(x, CORE_TRANS_MD, -protein->nodes[k].trans.MD, k + 1);
+    vit_set_core_trans(x, CORE_TRANS_IM, -protein->nodes[k].trans.IM, k + 1);
+    vit_set_core_trans(x, CORE_TRANS_II, -protein->nodes[k].trans.II, k + 0);
+    vit_set_core_trans(x, CORE_TRANS_DM, -protein->nodes[k].trans.DM, k + 1);
+    vit_set_core_trans(x, CORE_TRANS_DD, -protein->nodes[k].trans.DD, k + 1);
+  }
+  vit_set_core_trans(x, CORE_TRANS_MI, INFINITY, K - 1);
+  vit_set_core_trans(x, CORE_TRANS_II, INFINITY, K - 1);
+
+  for (size_t i = 0; i < VITERBI_TABLE_SIZE; ++i)
+  {
+    vit_set_null(x, -protein->null.emission[i], i);
+    vit_set_background(x, -protein->bg.emission[i], i);
+
+    for (int k = 0; k < K; ++k)
+      vit_set_match(x, -protein->nodes[k].emission[i], k, i);
+  }
+  return 0;
+}

c-core/setup_viterbi.h

@@ -0,0 +1,9 @@
+#ifndef SETUP_VITERBI_H
+#define SETUP_VITERBI_H
+
+struct vit;
+struct protein;
+
+int setup_viterbi(struct vit *, struct protein const *);
+
+#endif

c-core/test_protein.c

@@ -5,9 +5,11 @@
 #include "imm/nuclt_code.h"
 #include "imm/path.h"
 #include "protein.h"
+#include "setup_viterbi.h"
+#include "state.h"
+#include "trellis.h"
 #include "vendor/minctest.h"
-#include "viterbi.h"
-#include "viterbi_struct.h"
+#include "vit.h"
 
 static void test_protein_uniform(void);
 static void test_protein_occupancy(void);
@@ -19,6 +21,12 @@ int main(void)
   return lfails;
 }
 
+static int code_fn(int pos, int len, void *arg)
+{
+  struct imm_eseq const *seq = arg;
+  return imm_eseq_get(seq, pos, len, 1);
+}
+
 static void test_protein_uniform(void)
 {
   struct imm_path path = imm_path();
@@ -55,13 +63,14 @@ static void test_protein_uniform(void)
 
   eq(imm_eseq_setup(&eseq, &seq), 0);
 
-  struct viterbi task = {};
-  viterbi_init(&task);
-  viterbi_setup(&task, &protein, &eseq);
-  close(viterbi_null_loglik(&task), -48.9272687711);
-  float score = 0;
-  close(viterbi_alt_path(&task, &path, &score), 0);
-  close(score, -55.59428153448);
+  struct vit *task = vit_new();
+  eq(vit_setup(task, protein.core_size), 0);
+  eq(setup_viterbi(task, &protein), 0);
+  close(vit_null(task, imm_eseq_size(&eseq), code_fn, (void *)&eseq),  48.9272687711);
+  close(vit_cost(task, imm_eseq_size(&eseq), code_fn, (void *)&eseq), 55.59428153448);
+  eq(vit_path(task, imm_eseq_size(&eseq), code_fn, (void *)&eseq), 0);
+  imm_path_reset(&path);
+  eq(trellis_unzip(vit_trellis(task), imm_eseq_size(&eseq), &path), 0);
 
   eq(imm_path_nsteps(&path), 14);
 
@@ -100,7 +109,7 @@ static void test_protein_uniform(void)
   eq(i, 10);
 
   imm_eseq_cleanup(&eseq);
-  viterbi_cleanup(&task);
+  vit_del(task);
   protein_cleanup(&protein);
   imm_path_cleanup(&path);
 }
@@ -141,13 +150,14 @@ static void test_protein_occupancy(void)
 
   eq(imm_eseq_setup(&eseq, &seq), 0);
 
-  struct viterbi task = {};
-  viterbi_init(&task);
-  viterbi_setup(&task, &protein, &eseq);
-  close(viterbi_null_loglik(&task), -48.9272687711);
-  float score = 0;
-  close(viterbi_alt_path(&task, &path, &score), 0);
-  close(score, -54.35543421312);
+  struct vit *task = vit_new();
+  eq(vit_setup(task, protein.core_size), 0);
+  eq(setup_viterbi(task, &protein), 0);
+  close(vit_null(task, imm_eseq_size(&eseq), code_fn, (void *)&eseq), 48.9272687711);
+  close(vit_cost(task, imm_eseq_size(&eseq), code_fn, (void *)&eseq), 54.35543421312);
+  eq(vit_path(task, imm_eseq_size(&eseq), code_fn, (void *)&eseq), 0);
+  imm_path_reset(&path);
+  eq(trellis_unzip(vit_trellis(task), imm_eseq_size(&eseq), &path), 0);
 
   eq(imm_path_nsteps(&path), 14);
 
@@ -186,7 +196,7 @@ static void test_protein_occupancy(void)
   eq(i, 10);
 
   imm_eseq_cleanup(&eseq);
-  viterbi_cleanup(&task);
+  vit_del(task);
   protein_cleanup(&protein);
   imm_path_cleanup(&path);
 }

c-core/test_window.c

@@ -40,7 +40,7 @@ int main(void)
   eq(scan_add(scan, sequences[0].id, sequences[0].name, seq), 0);
   eq(scan_run(scan, PRODDIR), 0);
   eq(scan_progress(scan), 100);
-  eq(chksum(PRODDIR "/products.tsv"), 5268);
+  eq(chksum(PRODDIR "/products.tsv"), 5999);
   eq(scan_close(scan), 0);
 
   scan_del(scan);