REFINE module supports multiple labels

pbagos · May 22, 2022 · 4814641 · 4814641
1 parent 261b45f
commit 4814641
Show file tree

Hide file tree

Showing 22 changed files with 242 additions and 161 deletions.
diff --git a/.idea/.gitignore b/.idea/.gitignore
diff --git a/conf/conf.casino b/conf/conf.casino
@@ -53,6 +53,7 @@ STRDIV=9
 #Refine OPTIONS
 FLANK=3
 REFINE=false
+REFINE_LABELS=M
 ML_INIT=false
 
 # DECODING OPTIONS

diff --git a/conf/conf.cpg b/conf/conf.cpg
@@ -53,6 +53,7 @@ STRDIV=9
 #Refine OPTIONS
 FLANK=3
 REFINE=false
+REFINE_LABELS=M
 ML_INIT=false
 
 # DECODING OPTIONS

diff --git a/conf/conf.demo b/conf/conf.demo
@@ -17,7 +17,7 @@ EMISSIONS=FILE
 WEIGHTS=RPROP
 
 # Multithreaded parallelization for multicores
-PARALLEL=true
+PARALLEL=false
 defCPU=true
 nCPU=10
 
@@ -53,6 +53,7 @@ STRDIV=9
 #Refine OPTIONS
 FLANK=3
 REFINE=false
+REFINE_LABELS=M
 ML_INIT=false
 
 # DECODING OPTIONS
@@ -61,6 +62,7 @@ NBEST=false
 DYNAMIC=false
 POSVIT=false
 PLP=true
+LPB=false
 CONSTRAINT=false
 
 #EARLY STOPPING OPTIONS

diff --git a/conf/conf.hmmtm b/conf/conf.hmmtm
diff --git a/conf/conf.lipo b/conf/conf.lipo
@@ -53,6 +53,7 @@ STRDIV=9
 #Refine OPTIONS
 FLANK=3
 REFINE=false
+REFINE_LABELS=M
 ML_INIT=false
 
 # DECODING OPTIONS

diff --git a/conf/conf.signal b/conf/conf.signal
@@ -53,6 +53,7 @@ STRDIV=9
 #Refine OPTIONS
 FLANK=3
 REFINE=false
+REFINE_LABELS=M
 ML_INIT=false
 
 # DECODING OPTIONS

diff --git a/conf/conf.tat b/conf/conf.tat
@@ -53,6 +53,7 @@ STRDIV=9
 #Refine OPTIONS
 FLANK=3
 REFINE=false
+REFINE_LABELS=M
 ML_INIT=false
 
 # DECODING OPTIONS

diff --git a/conf/conf.tmbb b/conf/conf.tmbb
@@ -1,20 +1,20 @@
 # TRAINING OPTIONS
 RUN_CML=true
 RUN_GRADIENT=true
-HNN=true
+HNN=false
 ALLOW_BEGIN=true
 ALLOW_END=true
 RUN_ViterbiTraining=false
 threshold=0.02
-maxIter=50
+maxIter=200
 
-#PR
+#PROBABILITIES
 #FILE, RANDOM, UNIFORM, VITERBI
 TRANSITIONS=FILE
 #FILE, RANDOM, UNIFORM, VITERBI
 EMISSIONS=FILE
 #FILE, RANDOM_NORMAL, RANDOM_UNIFORM, RPROP, BOOT
-WEIGHTS=RPROP
+WEIGHTS=FILE
 
 # Multithreaded parallelization for multicores
 PARALLEL=false
@@ -53,6 +53,7 @@ STRDIV=9
 #Refine OPTIONS
 FLANK=3
 REFINE=true
+REFINE_LABELS=M
 ML_INIT=false
 
 # DECODING OPTIONS
@@ -61,7 +62,8 @@ NBEST=false
 DYNAMIC=false
 POSVIT=false
 PLP=true
-CONSTRAINT=false
+LPB=false
+CONSTRAINT=true
 
 #EARLY STOPPING OPTIONS
 EARLY=false
@@ -92,9 +94,9 @@ NOISE_EM=true
 PRIOR_TRANS=0.001
 
 #HNN OPTIONS
-windowLeft=5
-windowRight=5
-nhidden=9
+windowLeft=3
+windowRight=3
+nhidden=11
 ADD_GRAD=0.0
 DECAY=0.001
 #1: Sigmoid, 2: Sigmoid Modified, 3: Tanh

diff --git a/conf/conf.tmbbHNN b/conf/conf.tmbbHNN
@@ -17,7 +17,7 @@ EMISSIONS=FILE
 WEIGHTS=FILE
 
 # Multithreaded parallelization for multicores
-PARALLEL=false
+PARALLEL=true
 defCPU=true
 nCPU=10
 
@@ -53,6 +53,7 @@ STRDIV=9
 #Refine OPTIONS
 FLANK=3
 REFINE=true
+REFINE_LABELS=M
 ML_INIT=false
 
 # DECODING OPTIONS
@@ -61,6 +62,7 @@ NBEST=false
 DYNAMIC=false
 POSVIT=false
 PLP=true
+LPB=false
 CONSTRAINT=false
 
 #EARLY STOPPING OPTIONS
@@ -121,4 +123,4 @@ minDelta=1e-6
 etaInc=1.2
 etaDec=0.5
 
-#RANDOM SEQUENCE UTILITY
+#RANDOM SEQUENCE UTILITY
diff --git a/models/CpG.mdel b/models/CpG.mdel
@@ -1,7 +1,7 @@
 # MODEL OPTIONS
 MODEL=CpG
 
-ESYM=AGCT
+ESYM=ACGT
 OSYM=HLBE
 PSYM=HBE
 
@@ -14,15 +14,15 @@ outLabels=O
 # A+ C+ G+ T+ A- C- G- T-
 # A  C  G  T  a  c  g  t
 STATE =B A C G T a c g t E
-OSTATE=B A C G T a c g t E
-PSTATE=B H H H H H H H H E
+OSTATE=B H H H H L L L L E
+PSTATE=B H H H H L L L L E
 
 #MODEL PRIOR for every esym
 PRIOR = 0.077 0.018 0.058 0.066
 
 # Distribution for each osym
-# Each column must have a summ equal to 1
-# osym     H    B   E
-PRIOR1 = 0.97 0.0 0.0
-PRIOR2 = 0.01 0.0 0.0
-PRIOR3 = 0.02 0.0 0.0
+# Each column must have a sum equal to 1
+# osym     H    L   B   E
+PRIOR1 = 0.97 0.97 0.0 0.0
+PRIOR2 = 0.01 0.01 0.0 0.0
+PRIOR3 = 0.02 0.02 0.0 0.0
diff --git a/models/casino.mdel b/models/casino.mdel
@@ -19,7 +19,7 @@ PSTATE=B H H E
 PRIOR = 0.077 0.018 0.058 0.066 0.040 0.072
 
 # Distribution for each osym
-# Each column must have a summ equal to 1
+# Each column must have a sum equal to 1
 # osym     H    B   E
 PRIOR1 = 0.97 0.0 0.0
 PRIOR2 = 0.01 0.0 0.0

diff --git a/src/hmm/Args.java b/src/hmm/Args.java
@@ -3,7 +3,7 @@
 import java.io.*;
 
 public class Args {
-    static String Vesrion = "1.0.1";
+    static String Vesrion = "1.0.5";
 
     static boolean RUN_TRAINING = false;
     static boolean RUN_CROSSVAL = false;
@@ -54,7 +54,7 @@ public static final void Init(String[] args) {
             switch (a) {
                 case 'V':
 
-                    System.err.println("JUCHMME Version" + Vesrion);
+                    System.err.println("JUCHMME Version " + Vesrion);
                     System.exit(0);
 
                 case 'a':
@@ -140,6 +140,7 @@ public static final void Init(String[] args) {
                 case 't':
                     if (args.length > i + 1) {
                         file = args[i + 1];
+                        filesThree.add(args[i + 1]);
                         i++;
                         RUN_TRAINING = true;
                         Params.THREELINE = true;

diff --git a/src/hmm/Estimator.java b/src/hmm/Estimator.java
@@ -172,10 +172,18 @@ private HMM Estimate(SeqSet trainSet, Probs tab, WeightsL weightsL) throws Excep
                 //training.GetModel().print( new SystemOut() );
 
                 for (int j = 0; j < trainSet.nseqs; j++) {
-                    trainSet.seq[j].PutDashes(Model.transmLabels);
+                    System.out.println("Performed refinement " + trainSet.seq[j].header);
+                    //Refine multiple labels
+                    for (int i = 0; i < Params.REFINE_LABELS.length(); i++) {
+                        String label = Character.toString(Params.REFINE_LABELS.charAt(i));
+                        if (Model.psym.contains(label)) {
+                            trainSet.seq[j].PutDashes(label);
+                        } else {
+                            System.out.println("Warning: Refinement label "+label+" doesn't exist in model PSYM value.");
+                        }
+                    }
                     Viterbi vit2 = new Viterbi(training.GetModel(), trainSet.seq[j], false);
                     trainSet.seq[j].SetObs(vit2.getPath2());
-                    System.out.println("Performed refinement.");
                 }
             }
 

diff --git a/src/hmm/Juchmme.java b/src/hmm/Juchmme.java
@@ -24,7 +24,7 @@ public static void main(String[] args) throws Exception {
         long startTime = System.currentTimeMillis();
 
         System.out.println("JUCHMME :: Java Utility for Class Hidden Markov Models and Extensions");
-        System.out.println("Version 1.0.4; September 2019");
+        System.out.println("Version 1.0.5; September 2019");
         System.out.println("Copyright (C) 2019 Pantelis Bagos");
         System.out.println("Freely distributed under the GNU General Public Licence (GPLv3)");
         System.out.println("--------------------------------------------------------------------------");
@@ -109,6 +109,7 @@ public static void main(String[] args) throws Exception {
             if (Params.THREELINE) {
                 SeqSet testSet = new SeqSet((String) Args.filesThree.get(0));
                 Decoding dec = new Decoding(model, testSet, Params.CONSTRAINT, true, Params.parallel, Params.MSA);
+                stats.calcStats(testSet);
             }
 
             if (Params.FASTA) {