funcs.h

马尔科夫模型的java版本实现

/* function declarations */

/* readhmm */
int readhmm(FILE*, struct hmm_multi_s*);

/* readhmm_multialpha */
int readhmm_multialpha(FILE*, struct hmm_multi_s*);
void transform_singlehmmfile_to_multi(FILE *hmmfile, FILE *outfile);
int readhmm_check(FILE *hmmfile);
void copy_hmm_struct(struct hmm_multi_s *hmm, struct hmm_multi_s *retrain_hmm);

/* readseqs */
void get_sequences_std(FILE*, struct sequences_s*, struct hmm_s*);
void get_labeled_sequences_std(FILE*, struct sequences_s*, struct hmm_s*);
void get_sequences_fasta(FILE*, struct sequences_s*);
void get_sequences_msa_std(FILE*, FILE*, struct msa_sequences_s*, struct hmm_s*, int, struct replacement_letter_s*);
void get_sequences_msa_prf(FILE *seqfile, FILE *priorfile, struct msa_sequences_s *msa_seq_infop,
			   struct hmm_s *hmmp, int lead_seq);


/* readseqs_multi */
int seqfile_has_labels(FILE *seqfile);
void get_sequence_fasta_multi(FILE *seqfile, struct sequences_multi_s *seq_infop, int seq_nr);
void get_sequences_std_multi(FILE *seqfile, struct sequences_multi_s *seq_infop, struct hmm_multi_s *hmmp, int seq_nr);
void get_sequences_msa_std_multi(FILE*, FILE*, struct msa_sequences_multi_s*, struct hmm_multi_s*,
				 int, struct replacement_letter_multi_s*);
void get_sequences_msa_prf_multi(FILE *seqfile, FILE *priorfile, struct msa_sequences_multi_s *msa_seq_infop,
				 struct hmm_multi_s *hmmp);

/* savehmm */
int savehmm(FILE*, struct hmm_multi_s*);
int savehmm_multialpha(FILE*, struct hmm_multi_s*);


/* core_algorithms */
int forward(struct hmm_s*, struct letter_s*, struct forward_s**, double**, int);
int backward(struct hmm_s*, struct letter_s*, struct backward_s**, double*, int);
int viterbi(struct hmm_s*, struct letter_s*, struct viterbi_s**, int);
int one_best(struct hmm_s*, struct letter_s*, struct one_best_s**, double**, int, char*);
int msa_forward(struct hmm_s*, struct msa_sequences_s*, int,
		int, int, struct forward_s**, double**, int, int, int, double*);
int msa_backward(struct hmm_s*, struct msa_sequences_s*, int,
		int, struct backward_s**, double*, int, int, int, double*);
int msa_viterbi(struct hmm_s*, struct msa_sequences_s*, int,
		int, int, struct viterbi_s**, int, int, int, double*);
int msa_one_best(struct hmm_s*, struct msa_sequences_s*, int,
		int, int, struct one_best_s**, double**, int, char*, int, int, double*);


/* core_algorithms_multialpha */
int forward_multi(struct hmm_multi_s*, struct letter_s*,  struct letter_s*,  struct letter_s*,  struct letter_s*,
	    struct forward_s**, double**, int, int);
int backward_multi(struct hmm_multi_s*, struct letter_s*,  struct letter_s*,  struct letter_s*,  struct letter_s*,
	     struct backward_s**, double*, int, int);
int viterbi_multi(struct hmm_multi_s*, struct letter_s*,  struct letter_s*,  struct letter_s*,  struct letter_s*,
	    struct viterbi_s**, int, int);
int one_best_multi(struct hmm_multi_s*, struct letter_s*,  struct letter_s*,  struct letter_s*,  struct letter_s*,
	     struct one_best_s**, double**, int, char*, int);
int msa_forward_multi(struct hmm_multi_s*, struct msa_sequences_multi_s*, int,
		int, int, struct forward_s**, double**, int, int, int, int, double*, double*, double*, double*);
int msa_backward_multi(struct hmm_multi_s*, struct msa_sequences_multi_s*, int,
		       int, struct backward_s**, double*, int, int, int, int, double*, double*, double*, double*);
int msa_viterbi_multi(struct hmm_multi_s*, struct msa_sequences_multi_s*, int,
		int, int, struct viterbi_s**, int, int, int, int, double*, double*, double*, double*);
int msa_one_best_multi(struct hmm_multi_s*, struct msa_sequences_multi_s*, int,
		int, int, struct one_best_s**, double**, int, char*, int, int, int, double*, double*, double*, double*);

/* tm_core_algorithms */
int tm_viterbi(struct hmm_s*, struct letter_s*, struct viterbi_s**, struct aa_distrib_mtx_s*, int);


/* training_algorithms */
void baum_welch_std(struct hmm_s*, struct sequence_s*, int, int, int);
void baum_welch_dirichlet(struct hmm_s*, struct sequence_s*, int, int, int, int, int);
void extended_baum_welch_dirichlet(struct hmm_s*, struct sequence_s*, int, int, int, int, int);
void msa_baum_welch_dirichlet(struct hmm_s*, struct msa_sequences_s*, int, int, int, int, int, int, int, int, int, int, double*);
void extended_msa_baum_welch_dirichlet(struct hmm_s*, struct msa_sequences_s*, int, int, int, int, int, int, int, int, int, int,
				       double*);

/* training_algorithms */
void baum_welch_std_multi(struct hmm_multi_s *hmmp, struct sequence_multi_s *seqsp, int nr_seqs, int annealing, int use_labels,
			  int multi_scoring_method, int use_prior);
void baum_welch_dirichlet_multi(struct hmm_multi_s *hmmp, struct sequence_multi_s *seqsp, int nr_seqs, int annealing, int use_labels,
				int use_transition_pseudo_counts, int use_emission_pseudo_counts, int multi_scoring_method,
				int use_prior);
void msa_baum_welch_dirichlet_multi(struct hmm_multi_s *hmmp, struct msa_sequences_multi_s *msa_seq_infop, int nr_seqs,
				    int annealing,
				    int use_gap_shares, int use_lead_columns, int use_labels, int use_transition_pseudo_counts,
				    int use_emission_pseudo_counts, int normalize, int scoring_method, int use_nr_occ,
				    int multi_scoring_method, double *aa_freqs, double *aa_freqs_2, double *aa_freqs_3,
				    double *aa_freqs_4, int use_prior);
void extended_msa_baum_welch_dirichlet_multi(struct hmm_multi_s *hmmp, struct msa_sequences_multi_s *msa_seq_infop,
					     int nr_seqs, int annealing,
					     int use_gap_shares, int use_lead_columns, int use_labels,
					     int use_transition_pseudo_counts,
					     int use_emission_pseudo_counts, int normalize, int scoring_method, int use_nr_occ,
					     int multi_scoring_method, double *aa_freqs, double *aa_freqs_2, double *aa_freqs_3,
					     double *aa_freqs_4, int use_prior);



/* std_funcs */
void* malloc_or_die(int);
void init_float_mtx(double*, double, int);
void init_viterbi_s_mtx(struct viterbi_s*, double, int);
void printhelp_modhmms();
void printhelp_modhmms_msa();
void printhelp_hmmtrain();
void printhelp_hmmtrain_msa();
void printhelp_modhmms_multialpha();
void printhelp_modhmms_msa_multialpha();
void printhelp_hmmtrain_multialpha();
void printhelp_hmmtrain_msa_multialpha();
void printhelp_modhmms_tm_multialpha();
void printhelp_modhmms_tm_msa_multialpha();
void printhelp_hmmtrain_tm_multialpha();
void printhelp_hmmtrain_tm_msa_multialpha();
void printhelp_modhmms_tm();
void printhelp_modhmms_tm_msa();
void printhelp_hmmtrain_tm();
void printhelp_hmmtrain_tm_msa();

void printhelp_chmmtrain();
void printhelp_chmmtrain_msa();
void printhelp_chmmtrain_multialpha();
void printhelp_chmmtrain_msa_multialpha();
void printhelp_add_alphabet();
void printhelp_add2profilehmm();
void printhelp_cal();
void printhelp_opt();

int get_mtx_index(int,int,int);
int get_alphabet_index(struct letter_s*, char*, int);
int get_alphabet_index_msa_query(char*, char*, int);
int get_replacement_letter_index(struct letter_s*, struct replacement_letter_s*);
int get_replacement_letter_index_multi(struct letter_s *c, struct replacement_letter_multi_s *replacement_letters, int alphabet);
int get_alphabet_index_single(char*, char, int);
int get_replacement_letter_index_single(char*, struct replacement_letter_s*);
int get_seq_length(struct letter_s*);
int path_length(int, int, struct hmm_s*, int);
int path_length_multi(int, int, struct hmm_multi_s*, int);
void print_seq(struct letter_s*, FILE*, int, char*, int);
struct path_element* get_end_path_start(int l, struct hmm_s *hmmp);
struct path_element* get_end_path_start_multi(int l, struct hmm_multi_s *hmmp);
char* get_profile_vertex_type(int, int*);
void get_replacement_letters(FILE*, struct replacement_letter_s*);
void get_aa_distrib_mtx(FILE *distribmtxfile, struct aa_distrib_mtx_s *aa_distrib_mtxp);
void get_replacement_letters_multi(FILE *replfile, struct replacement_letter_multi_s *replacement_lettersp);
char* letter_as_string(struct letter_s*);