/* Function: esl_hyperexp_Read()
*
* Purpose: Reads hyperexponential parameters from an open <e>.
* which is an <ESL_FILEPARSER> tokenizer for an open stream.
*
* The first token is <K>, the number of mixture components.
* The second token is <mu>, the x offset shared by all components.
* Then for each mixture component <k=1..K>, it reads
* a mixture coefficient <q[k]> and a decay parameter
* <lambda[k]>.
*
* The <2K+2> data tokens must occur in this order, but
* they can be grouped into any number of lines, because the
* parser ignores line breaks.
*
* Anything after a <\#> character on a line is a comment, and
* is ignored.
*
* Returns: <eslOK> on success, and <ret_hxp> points to a new <ESL_HYPEREXP>
* object.
* <eslEFORMAT> on "normal" parse failure caused by a bad file
* format that's likely the user's fault.
*
* Throws: <eslEMEM> if allocation of the new <ESL_HYPEREXP> fails.
*
*
* FIXME: All our mixture models (esl_dirichlet, for example) should be
* reconciled w/ identical interfaces & behaviour.
*/
int
esl_hyperexp_Read(ESL_FILEPARSER *e, ESL_HYPEREXP **ret_hxp)
{
ESL_HYPEREXP *hxp = NULL;
char *tok;
int status = eslOK;
int nc;
int k;
double sum;
esl_fileparser_SetCommentChar(e, '#');
if ((status = esl_fileparser_GetToken(e, &tok, NULL)) != eslOK) goto ERROR;
nc = atoi(tok);
if (nc < 1) {
sprintf(e->errbuf, "Expected # of components K >= 1 as first token");
goto ERROR;
}
if ((hxp = esl_hyperexp_Create(nc)) == NULL) return eslEMEM; /* percolation */
if ((status = esl_fileparser_GetToken(e, &tok, NULL)) != eslOK) goto ERROR;
hxp->mu = atof(tok);
for (k = 0; k < hxp->K; k++)
{
if ((status = esl_fileparser_GetToken(e, &tok, NULL)) != eslOK) goto ERROR;
hxp->q[k] = atof(tok);
if ((status = esl_fileparser_GetToken(e, &tok, NULL)) != eslOK) goto ERROR;
hxp->lambda[k] = atof(tok);
if (hxp->q[k] < 0. || hxp->q[k] > 1.) {
sprintf(e->errbuf, "Expected a mixture coefficient q[k], 0<=q[k]<=1");
goto ERROR;
}
if (hxp->lambda[k] <= 0.) {
sprintf(e->errbuf, "Expected a lambda parameter, lambda>0");
goto ERROR;
}
}
sum = esl_vec_DSum(hxp->q, hxp->K);
if (fabs(sum-1.0) > 0.05) {
sprintf(e->errbuf, "Expected mixture coefficients to sum to 1");
goto ERROR;
}
esl_vec_DNorm(hxp->q, hxp->K);
*ret_hxp = hxp;
return eslOK;
ERROR:
esl_hyperexp_Destroy(hxp);
return eslEFORMAT;
}
/* read_mask_file
*
* Given an open file pointer, read the first token of the
* file and return it as *ret_mask. It must contain only
* '0' or '1' characters.
*
* Returns: eslOK on success.
*/
int
read_mask_file(char *filename, char *errbuf, char **ret_mask, int *ret_mask_len)
{
ESL_FILEPARSER *efp = NULL;
char *mask = NULL;
char *tok;
int toklen;
int n;
int status;
if (esl_fileparser_Open(filename, NULL, &efp) != eslOK) ESL_XFAIL(eslFAIL, errbuf, "failed to open %s in read_mask_file\n", filename);
esl_fileparser_SetCommentChar(efp, '#');
if((status = esl_fileparser_GetToken(efp, &tok, &toklen)) != eslOK) ESL_XFAIL(eslFAIL, errbuf, "failed to read a single token from %s\n", filename);
ESL_ALLOC(mask, sizeof(char) * (toklen+1));
for(n = 0; n < toklen; n++) {
if((tok[n] == '0') || (tok[n] == '1')) {
mask[n] = tok[n];
}
else { ESL_XFAIL(eslFAIL, errbuf, "read a non-0 and non-1 character (%c) in the mask file %s\n", tok[n], filename); }
}
mask[n] = '\0';
*ret_mask = mask;
*ret_mask_len = n;
esl_fileparser_Close(efp);
return eslOK;
ERROR:
if (efp) esl_fileparser_Close(efp);
if (mask) free(mask);
return status;
}
/* Function: esl_paml_ReadE()
* Incept: SRE, Fri Jul 9 09:27:24 2004 [St. Louis]
*
* Purpose: Read an amino acid rate matrix in PAML format from stream
* <fp>. Return it in two pieces: the symmetric E
* exchangeability matrix in <E>, and the stationary
* probability vector $\pi$ in <pi>.
* Caller provides the memory for both <E> and <pi>. <E>
* is a $20 \times 20$ matrix allocated as
* <esl_dmatrix_Create(20, 20)>. <pi> is an array with
* space for at least 20 doubles.
*
* The <E> matrix is symmetric for off-diagonal elements:
* $E_{ij} = E_{ij}$ for $i \neq j$. The on-diagonal
* elements $E_{ii}$ are not valid and should not be
* accessed. (They are set to zero.)
* The rate matrix will later be obtained from <E>
* and <pi> as
* $Q_{ij} = E_{ij} \pi_j$ for $i \neq j$
* and
* $Q_{ii} = -\sum_{j \neq i} Q_{ij}$
* then scaled to units of one
* substitution/site; see <esl_ratemx_E2Q()> and
* <esl_ratemx_ScaleTo()>.
*
* Data file format: First 190 numbers are a
* lower-triangular matrix E of amino acid
* exchangeabilities $E_{ij}$. Next 20 numbers are the
* amino acid frequencies $\pi_i$. Remainder of the
* datafile is ignored.
*
* The alphabet order in the matrix and the frequency
* vector is assumed to be "ARNDCQEGHILKMFPSTWYV"
* (alphabetical by three-letter code), which appears to be
* PAML's default order. This is transformed to Easel's
* "ACDEFGHIKLMNPQRSTVWY" (alphabetical by one-letter code)
* in the $E_{ij}$ and $\pi_i$ that are returned.
*
* Args: fp - open datafile for reading.
* E - RETURN: E matrix of amino acid exchangeabilities e_ij,
* symmetric (E_ij = E_ji),
* in Easel amino acid alphabet order A..Y.
* Caller provides appropriately allocated space.
* pi - RETURN: \pi_i vector of amino acid frequencies,
* in Easel amino acid alphabet order A..Y.
* Caller provides appropriately allocated space.
*
* Returns: <eslOK> on success.
* Returns <eslEOF> on premature end of file (parse failed), in which
* case the contents of <E> and <pi> are undefined.
*
* Throws: <eslEMEM> on internal allocation failure,
* and the contents of <E> and <pi> are undefined.
*
* Xref: STL8/p.56.
*/
int
esl_paml_ReadE(FILE *fp, ESL_DMATRIX *E, double *pi)
{
int status;
ESL_FILEPARSER *efp = NULL;
char *tok;
int i,j;
char *pamlorder = "ARNDCQEGHILKMFPSTWYV";
char *eslorder = "ACDEFGHIKLMNPQRSTVWY";
int perm[20];
if ((status = esl_dmatrix_SetZero(E)) != eslOK) goto ERROR;
esl_vec_DSet(pi, 20, 0.);
if ((efp = esl_fileparser_Create(fp)) == NULL) goto ERROR;
if ((status = esl_fileparser_SetCommentChar(efp, '#')) != eslOK) goto ERROR;
/* Construct the alphabet permutation we need.
* perm[i] -> original row/column i goes to row/column perm[i]
*/
for (i = 0; i < 20; i++)
perm[i] = (int) (strchr(eslorder, pamlorder[i]) - eslorder);
/* Read the s_ij matrix data in, permuting as we go. */
for (i = 1; i < 20; i++)
for (j = 0; j < i; j++)
{
if ((status = esl_fileparser_GetToken(efp, &tok, NULL)) != eslOK) goto ERROR;
E->mx[perm[i]][perm[j]] = atof(tok);
E->mx[perm[j]][perm[i]] = E->mx[perm[i]][perm[j]];
}
/* Read the pi_i vector in, permuting as we read. */
for (i = 0; i < 20; i++)
{
if ((status = esl_fileparser_GetToken(efp, &tok, NULL)) != eslOK) goto ERROR;
pi[perm[i]] = atof(tok);
}
esl_fileparser_Destroy(efp);
return eslOK;
ERROR:
if (efp != NULL) esl_fileparser_Destroy(efp);
return status;
}
/* Function: p7_bg_Read()
* Synopsis: Read background frequencies from a file.
*
* Purpose: Read new background frequencies from file <bgfile>,
* overwriting the frequencies previously in the
* <P7_BG> object <bg>.
*
* Note that <bg> is already created by the caller, not
* created here. Also note that <p7_bg_Read()> only reads
* residue background frequencies used for the "null
* model", whereas a <P7_BG> object contains additional
* information for the bias filter and for the biased
* composition correction.
*
* Args: bgfile - file to read.
* bg - existing <P7_BG> object provided by the caller.
* errbuf - OPTIONAL: space for an error message, upon parse errors; or NULL.
*
* Returns: <eslOK> on success, and background frequencies in <bg>
* are overwritten.
*
* <eslENOTFOUND> if <bgfile> can't be opened for reading.
* <eslEFORMAT> if parsing of <bgfile> fails for some
* reason. In both cases, <errbuf> contains a
* user-directed error message upon return, including (if
* relevant) the file name <bgfile> and the line number on
* which an error was detected. <bg> is unmodified.
*
* Throws: <eslEMEM> on allocation failure; <bg> is unmodified,
* and <errbuf> is empty.
*/
int
p7_bg_Read(char *bgfile, P7_BG *bg, char *errbuf)
{
ESL_FILEPARSER *efp = NULL;
float *fq = NULL;
int n = 0;
char *tok;
int toklen;
int alphatype;
ESL_DSQ x;
int status;
if (errbuf) errbuf[0] = '\0';
status = esl_fileparser_Open(bgfile, NULL, &efp);
if (status == eslENOTFOUND) ESL_XFAIL(eslENOTFOUND, errbuf, "couldn't open bg file %s for reading", bgfile);
else if (status != eslOK) goto ERROR;
esl_fileparser_SetCommentChar(efp, '#');
/* First token is alphabet type: amino | DNA | RNA */
status = esl_fileparser_GetToken(efp, &tok, &toklen);
if (status == eslEOF) ESL_XFAIL(eslEFORMAT, errbuf, "premature end of file [line %d of bgfile %s]", efp->linenumber, bgfile);
else if (status != eslOK) goto ERROR;
alphatype = esl_abc_EncodeType(tok);
if (alphatype == eslUNKNOWN) ESL_XFAIL(eslEFORMAT, errbuf, "expected alphabet type but saw \"%s\" [line %d of bgfile %s]", tok, efp->linenumber, bgfile);
else if (alphatype != bg->abc->type) ESL_XFAIL(eslEFORMAT, errbuf, "bg file's alphabet is %s; expected %s [line %d, %s]", tok, esl_abc_DecodeType(bg->abc->type), efp->linenumber, bgfile);
ESL_ALLOC(fq, sizeof(float) * bg->abc->K);
esl_vec_FSet(fq, bg->abc->K, -1.0);
while ((status = esl_fileparser_NextLine(efp)) == eslOK)
{
status = esl_fileparser_GetTokenOnLine(efp, &tok, &toklen);
if (status == eslEOL) ESL_XFAIL(eslEFORMAT, errbuf, "premature end of file [line %d of bgfile %s", efp->linenumber, bgfile);
else if (status != eslOK) goto ERROR;
if (toklen != 1 || ! esl_abc_CIsCanonical(bg->abc, *tok))
ESL_XFAIL(eslEFORMAT, errbuf, "expected to parse a residue letter; saw %s [line %d of bgfile %s]", tok, efp->linenumber, bgfile);
x = esl_abc_DigitizeSymbol(bg->abc, *tok);
if (fq[x] != -1.0) ESL_XFAIL(eslEFORMAT, errbuf, "already parsed probability of %c [line %d of bgfile %s]", bg->abc->sym[x], efp->linenumber, bgfile);
n++;
status = esl_fileparser_GetTokenOnLine(efp, &tok, &toklen);
if (status == eslEOL) ESL_XFAIL(eslEFORMAT, errbuf, "premature end of file, expected a probability [line %d of bgfile %s]", efp->linenumber, bgfile);
else if (status != eslOK) goto ERROR;
if (! esl_str_IsReal(tok)) ESL_XFAIL(eslEFORMAT, errbuf, "expected a probability, saw %s [line %d of bgfile %s]", tok, efp->linenumber, bgfile);
fq[x] = atof(tok);
status = esl_fileparser_GetTokenOnLine(efp, &tok, &toklen);
if (status == eslOK) ESL_XFAIL(eslEFORMAT, errbuf, "extra unexpected data found [line %d of bgfile %s]", efp->linenumber, bgfile);
else if (status != eslEOL) goto ERROR;
}
if (status != eslEOF) goto ERROR;
if ( n != bg->abc->K)
ESL_XFAIL(eslEFORMAT, errbuf, "expected %d residue frequencies, but found %d in bgfile %s", bg->abc->K, n, bgfile);
if ( esl_FCompare(esl_vec_FSum(fq, bg->abc->K), 1.0, 0.001) != eslOK)
ESL_XFAIL(eslEFORMAT, errbuf, "residue frequencies do not sum to 1.0 in bgfile %s", bgfile);
/* all checking complete. no more error cases. overwrite bg with the new frequencies */
esl_vec_FNorm(fq, bg->abc->K);
esl_vec_FCopy(fq, bg->abc->K, bg->f);
free(fq);
esl_fileparser_Close(efp);
return eslOK;
ERROR:
if (fq) free(fq);
if (efp) esl_fileparser_Close(efp);
return status;
}
