/*****************************************************************************
* MEME > training_set > alphabet > letter
* Read in a identifier and symbol pair for a letter.
****************************************************************************/
void mxml_alphabet_letter(void *ctx, char *id, char symbol, char* aliases, char complement, char *equals, char *name, int colour) {
CTX_T *data;
char sym[2];
data = (CTX_T*)ctx;
if (data->alph != NULL) {
// check that the symbol exists in the alphabet
if (!alph_is_concrete(data->alph, symbol)) {
local_error(data, "The symbol %c does not exist in the built-in alphabet %s.\n",
symbol, alph_name(data->alph));
return;
}
// Note: don't bother checking complement value as it will not be set
} else {
if (equals == NULL) {
alph_reader_core(data->alph_rdr, symbol, aliases, name, colour, complement);
} else {
alph_reader_ambig(data->alph_rdr, symbol, aliases, name, colour, equals);
}
}
// create a mapping from the id to the symbol
sym[0] = symbol;
sym[1] = '\0';
if (!rbtree_make(data->letter_lookup, id, sym)) {
local_error(data, "The letter identifier %s has been used before.\n", id);
}
}
/****************************************************************************
* Does a column of an alignment contain any ambiguity codes?
****************************************************************************/
BOOLEAN_T alignment_site_ambiguous(ALPH_T alph, int index, ALIGNMENT_T* alignment) {
int i;
for (i = 0; i < alignment->num_sequences; i++) {
char c = get_seq_char(index, alignment->sequences[i]);
if (!alph_is_concrete(alph, c)) return(TRUE);
}
return(FALSE);
}
/*
* plib_name name of prior library file
* desired_beta > 0, scale \beta_{i,j} so
* \sum_{i=0}^L \lambda_i \sum_{j=1}^20 \beta_{i,j}
* has this value
* == 0, don't scale prior
* < 0, just get alphabet
*/
PriorLib *read_PriorLib(char *plib_name, double desired_beta, ALPH_T *custom_alph)
{
int i,j, line=0;
int l;
PriorLib *temp;
char input[MAXS], foo[MAXS], alphabet[MAXALPH+1], checkstr[81], *token;
double x;
FILE *fp;
ALPH_T *alph;
// tlb
fp = fopen(plib_name, "r");
if (!fp) {
fprintf(stderr, "Can't find prior library %s\n", plib_name);
exit(1);
}
token = "Alphabet="; line++;
test_value(fscanf(fp,"%s %s\n", checkstr, alphabet) == 2);
if (strcmp(checkstr, token)) {
fprintf(stderr, "Line %d of prior library file \n %s \n"
"should start with \"%s\" "
"but it starts with \"%s\".\n", line, plib_name, token, checkstr);
exit(1);
}
// determine alphabet
if (custom_alph == NULL) {
alph = alph_type(alphabet, 30);
if (alph == NULL) {
fprintf(stderr, "The partial alphabet specified in the prior library file"
" does not match a built-in alphabet and no complete alphabet was specified.\n");
exit(1);
}
} else {
int alen_core;
alen_core = strlen(alphabet);
i = 0;
if (alen_core == alph_size_core(custom_alph)) {
for (i = 0; i < alen_core; i++) {
if (!alph_is_concrete(custom_alph, alphabet[i])) break;
}
}
if (i == 0 || i < alen_core) {
fprintf(stderr, "The partial alphabet specified in the prior library file"
" does not match the complete alphabet specified.\n");
exit(1);
}
alph = alph_hold(custom_alph);
}
token = "NumDistr="; line++;
test_value(fscanf(fp,"%s %d\n", checkstr, &l) == 2);
if (strcmp(checkstr, token)) {
fprintf(stderr, "Line %d of prior library file \n %s \n"
"should start with \"%s\" "
"but it starts with \"%s\"\n.", line, plib_name, token, checkstr);
exit(1);
}
temp = alloc_PriorLib(l, alph);
if (desired_beta < 0) {
fclose(fp);
return(temp);
}
for (i = 0; i < temp->L; i++) {
// Get rid of number=
ignore_value(fscanf(fp,"%*s %*s\n"));
// Mixture
ignore_value(fscanf(fp,"%*s"));
test_value(fscanf(fp,"%lf\n", &x) == 1);
temp->Mix[i] = x;
// B (strength)
ignore_value(fscanf(fp,"%*s"));
test_value(fscanf(fp,"%lf\n", &x) == 1);
temp->B[i] = x;
// Alpha
temp->Distr[i][0] = temp->B[i];
ignore_value(fscanf(fp,"%*s"));
for (j = 1; j < alph_size_wild(alph); j++) {
test_value(fscanf(fp,"%lg", &x) == 1);
temp->Distr[i][j] = x * temp->B[i];
}
// FullUpdate
ignore_value(fscanf(fp,"%*s"));
test_value(fscanf(fp,"%d\n", &(temp->FullUpdate[i])) == 1);
// QUpdate
ignore_value(fscanf(fp,"%*s"));
test_value(fscanf(fp,"%d\n", &(temp->QUpdate[i])) == 1);
// StructID
test_value(fgets(input, MAXS, fp) != NULL);
test_value(sscanf(input,"%s",foo) == 1);
input[strlen(input)-1] = '\0';
strcpy( (temp->StructID[i]), (input + strlen(foo)) );
// Comments
test_value(fgets(input, MAXS, fp) != NULL);
test_value(sscanf(input,"%s",foo) == 1);
strcpy( (temp->Comment[i]), (input + strlen(foo)) );
}
// tlb; scale beta to desired value
if (desired_beta > 0) {
int i, j;
double beta = 0;
double scale;
for (i=0; i<temp->L; i++) {
beta += temp->Mix[i] * temp->B[i];
}
/*printf("beta = %10.6f\n", beta);*/
scale = desired_beta/beta;
for (i=0; i<temp->L; i++) {
for (j=0; j < alph_size_wild(alph); j++) {
temp->Distr[i][j] *= scale;
}
}
}
fclose(fp);
alph_release(alph);
return(temp);
}
