void dxml_start_motif(void *ctx, char *id, char *seq, int length,
long num_sites, long p_hits, long n_hits,
double pvalue, double evalue, double uevalue) {
CTX_T *data;
MOTIF_T *motif;
data = (CTX_T*)ctx;
data->motif = (MOTIF_T*)mm_malloc(sizeof(MOTIF_T));
motif = data->motif;
memset(motif, 0, sizeof(MOTIF_T));
set_motif_id(seq, strlen(seq), motif);
set_motif_id2("", 0, motif);
set_motif_strand('+', motif);
motif->length = length;
motif->num_sites = num_sites;
motif->evalue = evalue;
// both DNA and RNA have 4 letters
motif->alph = data->fscope.alphabet;
motif->flags = MOTIF_BOTH_STRANDS; // DREME does not support the concept of single strand scanning (yet)
// allocate the matrix
motif->freqs = allocate_matrix(motif->length, alph_size(motif->alph, ALPH_SIZE));
motif->scores = NULL; // no scores in DREME xml
// no url in DREME
motif->url = strdup("");
// set by postprocessing
motif->complexity = -1;
motif->trim_left = 0;
motif->trim_right = 0;
}
/*****************************************************************************
* MEME > motifs > motif
* Construct the skeleton of a motif.
****************************************************************************/
void mxml_start_motif(void *ctx, char *id, char *name, char *alt, int width, double sites,
double llr, double ic, double re, double bayes_threshold,
double log10_evalue, double elapsed_time, char *url) {
CTX_T *data;
MOTIF_T *motif;
data = (CTX_T*)ctx;
data->mscope.motif = mm_malloc(sizeof(MOTIF_T));
motif = data->mscope.motif;
memset(motif, 0, sizeof(MOTIF_T));
set_motif_id(name, strlen(name), motif);
set_motif_id2(alt, sizeof(alt), motif);
set_motif_strand('+', motif);
motif->length = width;
motif->num_sites = sites;
motif->url = strdup(url);
motif->log_evalue = log10_evalue;
motif->evalue = pow(10.0, log10_evalue);
// calculate alphabet size
motif->alph = alph_hold(data->alph);
motif->flags = (data->fscope.strands == 2 ? MOTIF_BOTH_STRANDS : 0);
// allocate matricies
motif->freqs = allocate_matrix(motif->length, alph_size_core(motif->alph));
init_matrix(-1, motif->freqs);
motif->scores = allocate_matrix(motif->length, alph_size_core(motif->alph));
init_matrix(NO_SCORE, motif->scores);
// should be set by a post processing method
motif->complexity = -1;
motif->trim_left = 0;
motif->trim_right = 0;
// cache motif position
if (data->options & SCANNED_SITES) {
rbtree_put(data->motif_lookup, id, &(data->current_motif));
}
}
void read_regexp_file(
char* filename, // Name of MEME file IN
int* num_motifs, // Number of motifs retrieved OUT
MOTIF_T* motifs // The retrieved motifs - NOT ALLOCATED!
) {
FILE* motif_file; // MEME file containing the motifs.
char motif_name[MAX_MOTIF_ID_LENGTH+1];
char motif_regexp[MAX_MOTIF_WIDTH];
ARRAY_T* these_freqs;
MOTIF_T* m;
int i;
//Set things to the defaults.
*num_motifs = 0;
// Open the given MEME file.
if (open_file(filename, "r", TRUE, "motif", "motifs", &motif_file) == 0)
exit(1);
//Set alphabet - ONLY supports dna.
set_alphabet(verbosity, "ACGT");
while (fscanf(motif_file, "%s\t%s", motif_name, motif_regexp) == 2) {
/*
* Now we:
* 1. Fill in new motif (preallocated)
* 2. Assign name
* 3. Convert regexp into frequency table.
*/
m = &(motifs[*num_motifs]);
set_motif_id(motif_name, m);
m->length = strlen(motif_regexp);
/* Store the alphabet size in the motif. */
m->alph_size = get_alph_size(ALPH_SIZE);
m->ambigs = get_alph_size(AMBIG_SIZE);
/* Allocate memory for the matrix. */
m->freqs = allocate_matrix(m->length, get_alph_size(ALL_SIZE));
//Set motif frequencies here.
for (i=0;i<strlen(motif_regexp);i++) {
switch(toupper(motif_regexp[i])) {
case 'A':
set_matrix_cell(i,alphabet_index('A',get_alphabet(TRUE)),1,m->freqs);
break;
case 'C':
set_matrix_cell(i,alphabet_index('C',get_alphabet(TRUE)),1,m->freqs);
break;
case 'G':
set_matrix_cell(i,alphabet_index('G',get_alphabet(TRUE)),1,m->freqs);
break;
case 'T':
set_matrix_cell(i,alphabet_index('T',get_alphabet(TRUE)),1,m->freqs);
break;
case 'U':
set_matrix_cell(i,alphabet_index('U',get_alphabet(TRUE)),1,m->freqs);
break;
case 'R': //purines
set_matrix_cell(i,alphabet_index('G',get_alphabet(TRUE)),1,m->freqs);
set_matrix_cell(i,alphabet_index('A',get_alphabet(TRUE)),1,m->freqs);
break;
case 'Y': //pyramidines
set_matrix_cell(i,alphabet_index('T',get_alphabet(TRUE)),1,m->freqs);
set_matrix_cell(i,alphabet_index('C',get_alphabet(TRUE)),1,m->freqs);
break;
case 'K': //keto
set_matrix_cell(i,alphabet_index('G',get_alphabet(TRUE)),1,m->freqs);
set_matrix_cell(i,alphabet_index('T',get_alphabet(TRUE)),1,m->freqs);
break;
case 'M': //amino
set_matrix_cell(i,alphabet_index('A',get_alphabet(TRUE)),1,m->freqs);
set_matrix_cell(i,alphabet_index('C',get_alphabet(TRUE)),1,m->freqs);
break;
case 'S': //strong
set_matrix_cell(i,alphabet_index('G',get_alphabet(TRUE)),1,m->freqs);
set_matrix_cell(i,alphabet_index('C',get_alphabet(TRUE)),1,m->freqs);
break;
case 'W': //weak
set_matrix_cell(i,alphabet_index('A',get_alphabet(TRUE)),1,m->freqs);
set_matrix_cell(i,alphabet_index('T',get_alphabet(TRUE)),1,m->freqs);
break;
case 'B':
set_matrix_cell(i,alphabet_index('G',get_alphabet(TRUE)),1,m->freqs);
set_matrix_cell(i,alphabet_index('T',get_alphabet(TRUE)),1,m->freqs);
set_matrix_cell(i,alphabet_index('C',get_alphabet(TRUE)),1,m->freqs);
break;
case 'D':
set_matrix_cell(i,alphabet_index('G',get_alphabet(TRUE)),1,m->freqs);
set_matrix_cell(i,alphabet_index('A',get_alphabet(TRUE)),1,m->freqs);
set_matrix_cell(i,alphabet_index('T',get_alphabet(TRUE)),1,m->freqs);
break;
case 'H':
set_matrix_cell(i,alphabet_index('A',get_alphabet(TRUE)),1,m->freqs);
set_matrix_cell(i,alphabet_index('C',get_alphabet(TRUE)),1,m->freqs);
set_matrix_cell(i,alphabet_index('T',get_alphabet(TRUE)),1,m->freqs);
break;
case 'V':
set_matrix_cell(i,alphabet_index('G',get_alphabet(TRUE)),1,m->freqs);
set_matrix_cell(i,alphabet_index('C',get_alphabet(TRUE)),1,m->freqs);
set_matrix_cell(i,alphabet_index('A',get_alphabet(TRUE)),1,m->freqs);
break;
case 'N':
set_matrix_cell(i,alphabet_index('A',get_alphabet(TRUE)),1,m->freqs);
set_matrix_cell(i,alphabet_index('C',get_alphabet(TRUE)),1,m->freqs);
set_matrix_cell(i,alphabet_index('G',get_alphabet(TRUE)),1,m->freqs);
set_matrix_cell(i,alphabet_index('T',get_alphabet(TRUE)),1,m->freqs);
break;
}
}
/* Compute values for ambiguous characters. */
for (i = 0; i < m->length; i++) {
these_freqs = get_matrix_row(i, m->freqs);
fill_in_ambiguous_chars(FALSE, these_freqs);
}
/* Compute and store the motif complexity. */
m->complexity = compute_motif_complexity(m);
//Move our pointer along to do the next motif.
(*num_motifs)++;
}
}
