|  | Example Program HMM Silent States Hidden Markov Model with silent states code exampleA tutorial about HMMs with silent states
 | 1 | #include <iostream> 
 |  | 2 | #include <fstream> 
 |  | 3 | #include <seqan/graph_algorithms.h> 
 |  | 4 | #include <seqan/basic/basic_logvalue.h> 
 |  | 5 | 
 |  | 6 | using namespace seqan; 
 |  | 7 | 
 |  | 8 | int main() { 
 | 
 | 9 | typedef LogProb<> TProbability; 
 |  | 10 | typedef Dna TAlphabet; 
 |  | 11 | typedef Size<TAlphabet>::Type TSize; 
 |  | 12 | typedef Graph<Hmm<TAlphabet, TProbability> > THmm; 
 |  | 13 | typedef VertexDescriptor<THmm>::Type TVertexDescriptor; 
 |  | 14 | typedef EdgeDescriptor<THmm>::Type TEdgeDescriptor; 
 |  | 15 | 
 |  | 16 | Dna dnaA = Dna('A'); 
 |  | 17 | Dna dnaC = Dna('C'); 
 |  | 18 | Dna dnaG = Dna('G'); 
 |  | 19 | Dna dnaT = Dna('T'); 
 |  | 20 | 
 |  | 21 | THmm hmm; 
 |  | 22 | 
 | 
 | 23 | TVertexDescriptor begState = addVertex(hmm); 
 |  | 24 | assignBeginState(hmm, begState); 
 |  | 25 | 
 | 
 | 26 | TVertexDescriptor emitState1 = addVertex(hmm); 
 |  | 27 | emissionProbability(hmm, emitState1, dnaA) = 0.0; 
 |  | 28 | emissionProbability(hmm, emitState1, dnaC) = 0.8; 
 |  | 29 | emissionProbability(hmm, emitState1, dnaG) = 0.1; 
 |  | 30 | emissionProbability(hmm, emitState1, dnaT) = 0.1; 
 |  | 31 | 
 |  | 32 | TVertexDescriptor emitState2 = addVertex(hmm); 
 |  | 33 | emissionProbability(hmm, emitState2, dnaA) = 0.0; 
 |  | 34 | emissionProbability(hmm, emitState2, dnaC) = 0.2; 
 |  | 35 | emissionProbability(hmm, emitState2, dnaG) = 0.2; 
 |  | 36 | emissionProbability(hmm, emitState2, dnaT) = 0.6; 
 |  | 37 | 
 |  | 38 | TVertexDescriptor emitState3 = addVertex(hmm); 
 |  | 39 | emissionProbability(hmm, emitState3, dnaA) = 0.7; 
 |  | 40 | emissionProbability(hmm, emitState3, dnaC) = 0.1; 
 |  | 41 | emissionProbability(hmm, emitState3, dnaG) = 0.1; 
 |  | 42 | emissionProbability(hmm, emitState3, dnaT) = 0.1; 
 |  | 43 | 
 |  | 44 | TVertexDescriptor emitState4 = addVertex(hmm); 
 |  | 45 | emissionProbability(hmm, emitState4, dnaA) = 0.25; 
 |  | 46 | emissionProbability(hmm, emitState4, dnaC) = 0.25; 
 |  | 47 | emissionProbability(hmm, emitState4, dnaG) = 0.25; 
 |  | 48 | emissionProbability(hmm, emitState4, dnaT) = 0.25; 
 |  | 49 | 
 | 
 | 50 | TVertexDescriptor delState1 = addVertex(hmm, true); 
 |  | 51 | TVertexDescriptor delState2 = addVertex(hmm, true); 
 |  | 52 | TVertexDescriptor delState3 = addVertex(hmm, true); 
 |  | 53 | TVertexDescriptor delState4 = addVertex(hmm, true); 
 |  | 54 | 
 | 
 | 55 | TVertexDescriptor eState = addVertex(hmm); 
 |  | 56 | assignEndState(hmm, eState); 
 |  | 57 | 
 | 
 | 58 | addEdge(hmm, begState, emitState1, 0.5); 
 |  | 59 | addEdge(hmm, begState, delState1, 0.5); 
 |  | 60 | addEdge(hmm, emitState1, emitState2, 0.5); 
 |  | 61 | addEdge(hmm, emitState1, delState2, 0.5); 
 |  | 62 | addEdge(hmm, delState1, emitState2, 0.5); 
 |  | 63 | addEdge(hmm, delState1, delState2, 0.5); 
 |  | 64 | addEdge(hmm, emitState2, emitState3, 0.5); 
 |  | 65 | addEdge(hmm, emitState2, delState3, 0.5); 
 |  | 66 | addEdge(hmm, delState2, emitState3, 0.5); 
 |  | 67 | addEdge(hmm, delState2, delState3, 0.5); 
 |  | 68 | addEdge(hmm, emitState3, emitState4, 0.5); 
 |  | 69 | addEdge(hmm, emitState3, delState4, 0.5); 
 |  | 70 | addEdge(hmm, delState3, emitState4, 0.5); 
 |  | 71 | addEdge(hmm, delState3, delState4, 0.5); 
 |  | 72 | addEdge(hmm, emitState4, eState, 1.0); 
 |  | 73 | addEdge(hmm, delState4, eState, 1.0); 
 |  | 74 | 
 | 
 | 75 | ::std::cout << hmm << ::std::endl; 
 |  | 76 | 
 | 
 | 77 | String<Dna> sequence = "CA"; 
 |  | 78 | String<TVertexDescriptor> path; 
 |  | 79 | TProbability p = viterbiAlgorithm(hmm, sequence, path); 
 |  | 80 | ::std::cout << "Viterbi algorithm" << ::std::endl; 
 |  | 81 | ::std::cout << "Probability of the best path: " << p << ::std::endl; 
 |  | 82 | ::std::cout << "Sequence: " << ::std::endl; 
 |  | 83 | for(TSize i = 0; i<length(sequence); ++i) ::std::cout << sequence[i] << ','; 
 |  | 84 | ::std::cout << ::std::endl; 
 |  | 85 | ::std::cout << "State path: " << ::std::endl; 
 |  | 86 | for(TSize i = 0; i<length(path); ++i) { 
 |  | 87 | ::std::cout << path[i]; 
 |  | 88 | if (isSilent(hmm, path[i])) ::std::cout << " (Silent)"; 
 |  | 89 | if (i < length(path) - 1) ::std::cout << ','; 
 |  | 90 | } 
 |  | 91 | ::std::cout << ::std::endl; 
 |  | 92 | 
 | 
 | 93 | ::std::cout << "Forward algorithm" << ::std::endl; 
 |  | 94 | p = forwardAlgorithm(hmm, sequence); 
 |  | 95 | ::std::cout << "Probability that the HMM generated the sequence: " << p << ::std::endl; 
 |  | 96 | 
 | 
 | 97 | ::std::cout << "Backward algorithm" << ::std::endl; 
 |  | 98 | p = backwardAlgorithm(hmm, sequence); 
 |  | 99 | ::std::cout << "Probability that the HMM generated the sequence: " << p << ::std::endl; 
 |  | 100 | 
 |  | 101 | return 0; 
 |  | 102 | } 
 | 
  |