funcs.h

马尔科夫模型的java版本实现

char* sequence_as_string(struct letter_s*);
void get_viterbi_label_path(struct viterbi_s *cur, struct hmm_s *hmmp,
			    struct viterbi_s *viterbi_mtxp, int row, int row_size, char *labels, int *ip);
void get_viterbi_label_path_multi(struct viterbi_s *cur, struct hmm_multi_s *hmmp,
				  struct viterbi_s *viterbi_mtxp, int row, int row_size, char *labels, int *ip);
void get_viterbi_path(struct viterbi_s *cur, struct hmm_s *hmmp,
		      struct viterbi_s *viterbi_mtxp, int row, int row_size, int *path, int *ip);
void get_viterbi_path_multi(struct viterbi_s *cur, struct hmm_multi_s *hmmp,
			    struct viterbi_s *viterbi_mtxp, int row, int row_size, int *path, int *ip);
void itoa(char* s, int nr);
void ftoa(char* s, double nr, int prec);
int read_subst_matrix(double **mtx, FILE *substmtxfile);
int read_subst_matrix_multi(double **mtxpp, double **mtxpp_2, double **mtxpp_3, double **mtxpp_4, FILE *substmtxfile);
int read_prior_file(struct emission_dirichlet_s *em_di, struct hmm_s *hmmp, FILE *priorfile);
int read_frequencies(FILE *freqfile, double **aa_freqs);
int read_frequencies_multi(FILE *freqfile, double **aa_freqsp, double **aa_freqsp_2, double **aa_freqsp_3, double **aa_freqsp_4);
int read_prior_file_multi(struct emission_dirichlet_s *em_di, struct hmm_multi_s *hmmp, FILE *priorfile, int alphabet);
int locked_state(struct hmm_s *hmmp, int v);
int locked_state_multi(struct hmm_multi_s *hmmp, int v);
int get_best_reliability_score(double reliability_score_1, double reliability_score_2, double reliability_score_3);
void hmm_garbage_collection(FILE *hmmfile, struct hmm_s *hmmp);
void hmm_garbage_collection_multi(FILE *hmmfile, struct hmm_multi_s *hmmp);
void hmm_garbage_collection_multi_no_dirichlet(FILE *hmmfile, struct hmm_multi_s *hmmp);
void msa_seq_garbage_collection_multi(struct msa_sequences_multi_s *msa_seq_info, int nr_alphabets);
void seq_garbage_collection_multi(struct sequences_multi_s *seq_info, int nr_alphabets);
void get_msa_labels(FILE *labelfile, struct msa_sequences_s *msa_seq_infop, struct hmm_s *hmmp);
void get_msa_labels_all_columns(FILE *labelfile, struct msa_sequences_s *msa_seq_infop, struct hmm_s *hmmp);
int update_shares_prior(struct emission_dirichlet_s *em_di, struct hmm_s *hmmp,
			struct msa_sequences_s *msa_seq_infop, int l);
int replacement_letter(struct letter_s *cur_letterp, struct replacement_letter_s *replacement_letters, 
		       struct msa_sequences_s *msa_seq_infop, struct hmm_s *hmmp, int seq_pos);
void get_labels_multi(FILE *labelfile, struct sequences_multi_s *seq_infop, struct hmm_multi_s *hmmp, int seq_nr);
void get_msa_labels_multi(FILE *labelfile, struct msa_sequences_multi_s *msa_seq_infop, struct hmm_multi_s *hmmp);
void get_msa_labels_all_columns_multi(FILE *labelfile, struct msa_sequences_multi_s *msa_seq_infop, struct hmm_multi_s *hmmp);
int update_shares_prior_multi(struct emission_dirichlet_s *em_di, struct hmm_multi_s *hmmp,
			struct msa_sequences_multi_s *msa_seq_infop, int l, int alphabet);
int replacement_letter_multi(struct letter_s *cur_letterp, struct replacement_letter_multi_s *replacement_letters, 
		       struct msa_sequences_multi_s *msa_seq_infop, struct hmm_multi_s *hmmp, int seq_pos, int alphabet);
int get_nr_alphabets(FILE *hmmfile);
void get_set_of_labels(struct hmm_s *hmmp);
void get_set_of_labels_multi(struct hmm_multi_s *hmmp);
void get_reverse_msa_seq_multi(struct msa_sequences_multi_s *msa_seq_infop, struct msa_sequences_multi_s *reverse_msa_seq_infop,
			       struct hmm_multi_s *hmmp);
void get_reverse_seq_multi(struct sequence_multi_s *seqs, struct letter_s **reverse_seq_1,
			   struct letter_s **reverse_seq_2, struct letter_s **reverse_seq_3,
			   struct letter_s **reverse_seq_4, struct hmm_multi_s *hmmp, int seq_len);

/* std calculation funcs */
double get_single_gaussian_statescore(double mu, double sigma_square, double letter);
double get_dp_statescore(int a_size, int use_gap_shares, int use_prior_shares, struct msa_letter_s *msa_seq,
			 int p, double *emissions,  int vertex, int normalize, double *gap_shares);
double get_dp_picasso_statescore(int a_size, int use_gap_shares, int use_prior_shares, struct msa_letter_s *msa_seq,
				 int p, double *emissions,  int vertex, int normalize, double *gap_shares, double *aa_freqs);
double get_sjolander_statescore(int a_size, int use_gap_shares, int use_prior_shares, struct msa_letter_s *msa_seq,
			       int p, double *emissions, int vertex, int normalize, double *gap_shares);
double get_sjolander_reversed_statescore(int a_size, int use_gap_shares, int use_prior_shares, struct msa_letter_s *msa_seq,
					 int p, double *emissions,  int vertex, int normalize, double *gap_shares);
double get_picasso_statescore(int a_size, int use_gap_shares, int use_prior_shares, struct msa_letter_s *msa_seq,
			 int p, double *emissions,  int vertex, int normalize, double *gap_shares, double *aa_freqs);
double get_picasso_sym_statescore(int a_size, int use_gap_shares, int use_prior_shares, struct msa_letter_s *msa_seq,
			 int p, double *emissions,  int vertex, int normalize, double *gap_shares, double *aa_freqs);
double get_subst_mtx_product_statescore(int a_size, int use_gap_shares, int use_prior_shares, struct msa_letter_s *msa_seq,
					int p, double *emissions, int vertex, double *subst_mtx);
double get_subst_mtx_dot_product_statescore(int a_size, int use_gap_shares, int use_prior_shares, struct msa_letter_s *msa_seq,
					    int p, double *emissions,  int vertex, int normalize, double *gap_shares,
					    int query_index, double *subst_mtx);
double get_subst_mtx_dot_product_prior_statescore(int a_size, int use_gap_shares, int use_prior_shares, struct msa_letter_s *msa_seq,
						  int p, double *emissions,  int vertex, int normalize, double *gap_shares,
						  int query_index, double *subst_mtx);

void add_to_E_dot_product(double *E, double Eka_base, struct msa_letter_s *msa_seq, int p,
			  int k, int a_size, int normalize);
void add_to_E_dot_product_picasso(double *E, double Eka_base, struct msa_letter_s *msa_seq, int p,
				  int k, int a_size, int normalize);
void add_to_E_picasso(double *E, double Eka_base, struct msa_letter_s *msa_seq, int p,
		      int k, int a_size, int normalize);
void add_to_E_picasso_sym(double *E, double Eka_base, struct msa_letter_s *msa_seq, int p,
		      int k, int a_size, int normalize);
void add_to_E_sjolander_score(double *E, double Eka_base, struct msa_letter_s *msa_seq, int p,
			      int k, int a_size, int normalize);
void add_to_E_sjolander_reversed_score(double *E, double Eka_base, struct msa_letter_s *msa_seq, int p,
				       int k, int a_size, int normalize);
void add_to_E_subst_mtx_product(double *E, double Eka_base, struct msa_letter_s *msa_seq, int p,
				int k, int a_size, int normalize, double *subst_mtx);
void add_to_E_subst_mtx_dot_product(double *E, double Eka_base, struct msa_letter_s *msa_seq, int p,
				    int k, int a_size, int normalize, double *subst_mtx, char *alphabet);
void add_to_E_subst_mtx_dot_product_prior(double *E, double Eka_base, struct msa_letter_s *msa_seq, int p,
					  int k, int a_size, int normalize, double *subst_mtx, char *alphabet);

void add_to_E_dot_product_nr_occ(double *E, double Eka_base, struct msa_letter_s *msa_seq, int p,
				int k, int a_size, int normalize);
void add_to_E_dot_product_picasso_nr_occ(double *E, double Eka_base, struct msa_letter_s *msa_seq, int p,
					 int k, int a_size, int normalize);
void add_to_E_picasso_nr_occ(double *E, double Eka_base, struct msa_letter_s *msa_seq, int p,
			     int k, int a_size, int normalize);
void add_to_E_picasso_sym_nr_occ(double *E, double Eka_base, struct msa_letter_s *msa_seq, int p,
				 int k, int a_size, int normalize);
void add_to_E_sjolander_score_nr_occ(double *E, double Eka_base, struct msa_letter_s *msa_seq, int p,
				int k, int a_size, int normalize);
void add_to_E_sjolander_reversed_score_nr_occ(double *E, double Eka_base, struct msa_letter_s *msa_seq, int p,
					      int k, int a_size, int normalize);
void add_to_E_subst_mtx_product_nr_occ(double *E, double Eka_base, struct msa_letter_s *msa_seq, int p,
				       int k, int a_size, int normalize, double *subst_mtx);
void add_to_E_subst_mtx_dot_product_nr_occ(double *E, double Eka_base, struct msa_letter_s *msa_seq, int p,
					   int k, int a_size, int normalize, double *subst_mtx, char *alphabet);
void add_to_E_subst_mtx_dot_product_prior_nr_occ(double *E, double Eka_base, struct msa_letter_s *msa_seq, int p,
					  int k, int a_size, int normalize, double *subst_mtx, char *alphabet);

void update_labelings(struct one_best_s *cur_rowp, char *vertex_labels, 
		      int *sorted_v_list, int seq_len, int c, char *labels, int nr_of_labels, int nr_v);
void deallocate_row_labelings(struct one_best_s *prev_rowp, int nr_v);


/* debug_funcs */
void dump_align_matrix(int nr_rows, int nr_cols, struct align_mtx_element_s *mtx);
void dump_trans_matrix(int,int,double*);
void dump_int_trans_matrix(int nr_rows, int nr_cols, double *mtx);
void dump_emiss_matrix(int,int,double*);
void dump_post_prob_matrix(int nr_rows, int nr_cols, double *mtx);
void dump_forward_matrix(int,int,struct forward_s*);
void dump_backward_matrix(int,int,struct backward_s*);
void dump_viterbi_matrix(int nr_rows, int nr_cols, struct viterbi_s *mtx);
void dump_one_best_matrix(int, int, struct one_best_s*);
void dump_scaling_array(int,double*);
void dump_from_trans_array(int,struct path_element**);
void dump_to_trans_array(int,struct path_element**);
void dump_viterbi_path(struct viterbi_s*, struct hmm_s*, struct viterbi_s*, int, int);
void dump_viterbi_label_path(struct viterbi_s*, struct hmm_s*, struct viterbi_s*, int, int);
void dump_T_matrix(int,int,double*);
void dump_E_matrix(int,int,double*);
void dump_modules(struct hmm_s*);
void dump_distrib_groups(int*, int);
void dump_trans_tie_groups(struct transition_s*, int);
void dump_prior_struct(struct emission_dirichlet_s*);
void dump_silent_vertices(struct hmm_s*);
void dump_silent_vertices_multi(struct hmm_multi_s *hmmp);
void dump_locked_vertices(struct hmm_s*);
void dump_seqs(struct sequences_s*);
void dump_seqs_multi(struct sequences_multi_s*);
void dump_msa_seqs(struct msa_sequences_s*, int);
void dump_msa_seqs_multi(struct msa_sequences_s*, struct hmm_multi_s*);
void dump_to_silent_trans_array(int, int**);
void dump_aa_distrib_mtx(struct aa_distrib_mtx_s *aa_distrib_mtxp);
void dump_v_list(int*);
void dump_labeling(char*, int);
void dump_label_tmp_list(int *list);
void check_for_corrupt_values(int nr_rows, int nr_cols, double *mtx, char *name);
void dump_subst_mtx(int a_size, double *mtx);
void dump_multi_modules(struct hmm_multi_s *hmmp);
void dump_weights(double *total_weights, int nr_seqs);