AjBool ajWriteline(AjPFile file, const AjPStr line)
{
if(!fwrite(MAJSTRGETPTR(line), MAJSTRGETLEN(line), 1, file->fp))
return ajFalse;
return ajTrue;
}
void ajXmlClear(AjPXml xml)
{
if(MAJSTRGETLEN(xml->Id))
ajStrSetClear(&xml->Id);
if(MAJSTRGETLEN(xml->Db))
ajStrSetClear(&xml->Db);
if(MAJSTRGETLEN(xml->Setdb))
ajStrSetClear(&xml->Setdb);
if(MAJSTRGETLEN(xml->Full))
ajStrSetClear(&xml->Full);
if(MAJSTRGETLEN(xml->Qry))
ajStrSetClear(&xml->Qry);
if(MAJSTRGETLEN(xml->Formatstr))
ajStrSetClear(&xml->Formatstr);
if(MAJSTRGETLEN(xml->Filename))
ajStrSetClear(&xml->Filename);
ajStrDel(&xml->TextPtr);
xml->Count = 0;
xml->Fpos = 0L;
xml->Format = 0;
if(xml->Doc)
ajDomDocumentDestroyNode(xml->Doc, &xml->Doc);
return;
}
AjBool ajWritelineNewline(AjPFile file, const AjPStr line)
{
if(!fwrite(MAJSTRGETPTR(line), 1, MAJSTRGETLEN(line), file->fp))
return ajFalse;
#ifdef WIN32
if(!fwrite("\r\n", 1, 2, file->fp))
return ajFalse;
#else
if(!fwrite("\n", 1, 1, file->fp))
return ajFalse;
#endif
return ajTrue;
}
void ajRefseqClear(AjPRefseq refseq)
{
AjPSeqRange tmprange = NULL;
if(MAJSTRGETLEN(refseq->Id))
ajStrSetClear(&refseq->Id);
if(MAJSTRGETLEN(refseq->Db))
ajStrSetClear(&refseq->Db);
if(MAJSTRGETLEN(refseq->Setdb))
ajStrSetClear(&refseq->Setdb);
if(MAJSTRGETLEN(refseq->Full))
ajStrSetClear(&refseq->Full);
if(MAJSTRGETLEN(refseq->Qry))
ajStrSetClear(&refseq->Qry);
if(MAJSTRGETLEN(refseq->Formatstr))
ajStrSetClear(&refseq->Formatstr);
if(MAJSTRGETLEN(refseq->Filename))
ajStrSetClear(&refseq->Filename);
ajStrDel(&refseq->TextPtr);
ajStrDel(&refseq->Desc);
ajStrDel(&refseq->Seq);
while(ajListPop(refseq->Seqlist,(void **)&tmprange))
ajSeqrangeDel(&tmprange);
ajListFree(&refseq->Seqlist);
refseq->Count = 0;
refseq->Fpos = 0L;
refseq->Format = 0;
return;
}
/* @funcstatic sigscanlig_WriteFastaHit ***************************************
**
** Write a Hit from a Hitlist object to an output file in embl-like format
** (see documentation for the DOMAINATRIX "seqsearch" application).
** Text for Class, Fold, Superfamily and Family is only written if the text
** is available.
**
** @param [u] outf [AjPFile] Output file stream
** @param [r] hits [const AjPList] List of hit objects.
** @param [r] n [ajint] Number of hit to generate.
** @param [r] DOSEQ [AjBool] True if sequence is to be printed.
**
** @return [AjBool] True on success
** @@
******************************************************************************/
AjBool sigscanlig_WriteFastaHit(AjPFile outf, AjPList hits, ajint n,
AjBool DOSEQ)
{
/* AjBool sigscanlig_WriteFastaHit(AjPFile outf, AjPList siglist,
AjPList hits, ajint n, AjBool DOSEQ) */
EmbPHit hit = NULL;
EmbPSignature sig = NULL;
/* EmbPSignature *sigarr = NULL; */
EmbPHit *hitarr = NULL;
ajint sizarr=0;
/* if(!outf || !siglist || !hits)
return ajFalse; */
if(!outf || !hits)
return ajFalse;
/* sizarr = ajListToarray(siglist, (void ***) &sigarr);
if(sizarr != ajListToarray(hits, (void ***) &hitarr))
ajFatal("Arrays are different sizes"); */
sizarr = ajListToarray(hits, (void ***) &hitarr);
if(n>=sizarr)
ajFatal("Requested hit out of range in sigscanlig_WriteFastaHit");
hit = hitarr[n];
sig = hitarr[n]->Sig;
/* sig = sigarr[n]; */
ajFmtPrintF(outf, "> ");
if(MAJSTRGETLEN(hit->Acc))
ajFmtPrintF(outf, "%S^", hit->Acc);
else
ajFmtPrintF(outf, ".^");
if(MAJSTRGETLEN(hit->Spr))
ajFmtPrintF(outf, "%S^", hit->Spr);
else
ajFmtPrintF(outf, ".^");
ajFmtPrintF(outf, "%d^%d^", hit->Start, hit->End);
ajFmtPrintF(outf, "LIGAND^");
if(MAJSTRGETLEN(sig->Id))
ajFmtPrintF(outf, "%S^", sig->Id);
else
ajFmtPrintF(outf, ".^");
if(MAJSTRGETLEN(sig->Domid))
ajFmtPrintF(outf, "%S^", sig->Domid);
else
ajFmtPrintF(outf, ".^");
if(MAJSTRGETLEN(sig->Ligid))
ajFmtPrintF(outf, "%S^", sig->Ligid);
else
ajFmtPrintF(outf, ".^");
ajFmtPrintF(outf,"%d^", sig->sn);
ajFmtPrintF(outf,"%d^", sig->ns);
ajFmtPrintF(outf,"%d^", sig->pn);
ajFmtPrintF(outf,"%d^", sig->np);
if(sig->Typesig == aj1D)
ajFmtPrintF(outf,"1D");
else if(sig->Typesig == aj3D)
ajFmtPrintF(outf,"3D");
else
ajFatal("Signature type unknown in sigscanlig_WriteFasta");
if(sig->np)
ajFmtPrintF(outf,"P^");
else
ajFmtPrintF(outf,"F^");
ajFmtPrintF(outf, "%.2f^", hit->Score);
ajFmtPrintF(outf, "%.3e^", hit->Pval);
ajFmtPrintF(outf, "%.3e", hit->Eval);
if(DOSEQ)
{
ajFmtPrintF(outf, "\n");
ajFmtPrintF(outf, "%S\n", hit->Seq);
}
/* AJFREE(sigarr); */
AJFREE(hitarr);
return ajTrue;
}
AjBool sigscanlig_WriteFasta(AjPFile outf, AjPList hits)
{
ajint x = 0;
EmbPHit hit = NULL;
EmbPSignature sig = NULL;
EmbPSignature *sigarr = NULL;
EmbPHit *hitarr = NULL;
ajint sizarr=0;
/*
if(!outf || !siglist || !hits)
return ajFalse;
*/
if(!outf || !hits)
return ajFalse;
/*
sizarr = ajListToarray(siglist, (void ***) &sigarr);
if(sizarr != ajListToarray(hits, (void ***) &hitarr))
ajFatal("Arrays are different sizes");
*/
sizarr = ajListToarray(hits, (void ***) &hitarr);
for(x=0; x<sizarr; x++)
{
/* There has to be a hit for each signature for correct
generation of the LHF by sigscanlig_WriteFasta. Therefore
empty hits may have been pushed. Catch those here. */
/* if(!MAJSTRGETLEN(hit->Model))
continue; */
hit = hitarr[x];
/* sig = sigarr[x]; */
sig = hitarr[x]->Sig;
ajFmtPrintF(outf, "> ");
if(MAJSTRGETLEN(hit->Acc))
ajFmtPrintF(outf, "%S^", hit->Acc);
else
ajFmtPrintF(outf, ".^");
if(MAJSTRGETLEN(hit->Spr))
ajFmtPrintF(outf, "%S^", hit->Spr);
else
ajFmtPrintF(outf, ".^");
ajFmtPrintF(outf, "%d^%d^", hit->Start, hit->End);
ajFmtPrintF(outf, "LIGAND^");
if(MAJSTRGETLEN(sig->Id))
ajFmtPrintF(outf, "%S^", sig->Id);
else
ajFmtPrintF(outf, ".^");
if(MAJSTRGETLEN(sig->Domid))
ajFmtPrintF(outf, "%S^", sig->Domid);
else
ajFmtPrintF(outf, ".^");
if(MAJSTRGETLEN(sig->Ligid))
ajFmtPrintF(outf, "%S^", sig->Ligid);
else
ajFmtPrintF(outf, ".^");
ajFmtPrintF(outf,"%d^", sig->sn);
ajFmtPrintF(outf,"%d^", sig->ns);
ajFmtPrintF(outf,"%d^", sig->pn);
ajFmtPrintF(outf,"%d^", sig->np);
if(sig->Typesig == aj1D)
ajFmtPrintF(outf,"1D");
else if(sig->Typesig == aj3D)
ajFmtPrintF(outf,"3D");
else
ajFatal("Signature type unknown in sigscanlig_WriteFasta");
if(sig->np)
ajFmtPrintF(outf,"P^");
else
ajFmtPrintF(outf,"F^");
ajFmtPrintF(outf, "%.2f^", hit->Score);
ajFmtPrintF(outf, "%.3e^", hit->Pval);
ajFmtPrintF(outf, "%.3e", hit->Eval);
ajFmtPrintF(outf, "\n");
ajFmtPrintF(outf, "%S\n", hit->Seq);
}
AJFREE(sigarr);
AJFREE(hitarr);
return ajTrue;
}
static AjBool sigscanlig_SignatureAlignWriteBlock(AjPFile outf,
AjPList hits)
/*
static AjBool sigscanlig_SignatureAlignWriteBlock(AjPFile outf,
AjPList siglist,
AjPList hits)
*/
{
/*
** A line of the alignment (including accession number, a space and the
** sequence) in the output file is 70 characters long. An index number is
** also printed after this 70 character field.
*/
ajint wid1 = 0; /*Temp. width of Accession Number */
ajint mwid1 = 0; /*
** Max. width of Accession Number or the string
** "Number".
** This is the field width the accession numbers
** will be printed into
*/
ajint mwid2 = 0; /* Width of region to print sequence into */
ajint len = 0; /* Temp. length of sequence */
ajint mlen = 0; /* Max. length of sequence */
const char *ptrp = NULL; /* Pointer to sequence string */
const char *ptrs = NULL; /* Pointer to alignment string */
ajint idx = 0; /* Start position for printing */
ajint niter = 0; /*
** No. iterations of loop for printing out
** sequence blocks
*/
ajint fwid1 = 70; /*
** Including accession number, a space, 7 characters
** for the first index number, and the sequence
*/
ajint fwid2 = 7; /* Field width for the first index number */
ajint num = 0; /* Index number for alignment */
ajint y = 0; /* Loop counter */
AjIList iter = NULL;
AjIList itersig = NULL;
EmbPHit hit = NULL;
ajint hitcnt=0; /* Counter for current hit */
AjPStr label = NULL;
/* Check args */
/* if(!outf || !hits || !siglist)
return ajFalse; */
if(!outf || !hits)
return ajFalse;
/* Cycle through hits to find longest width of accession number. */
len=0;
mlen=0;
wid1=0;
mwid1=0;
label = ajStrNew();
iter = ajListIterNew(hits);
while((hit = (EmbPHit) ajListIterGet(iter)))
{
if((wid1=MAJSTRGETLEN(hit->Sig->Ligid))>mwid1)
mwid1 = wid1;
if((len=MAJSTRGETLEN(hit->Seq))>mlen)
mlen = len;
}
ajListIterDel(&iter);
ajListIterDel(&itersig);
/* Assign field widths and number of iterations for printing. */
if((wid1 = strlen("SIGNATURE"))>mwid1)
mwid1 = wid1;
mwid1++; /* A space */
mwid1 += 8; /* 2 underscores + 2 integers in 3 char spacing each. */
mwid2 = fwid1-fwid2-mwid1;
niter = (ajint)ceil( ((double)mlen/(double)mwid2));
/* Print header info and SCOP classification records of signature */
ajFmtPrintF(outf,
"# DE Alignment of query sequence against library of "
"signatures\n");
/* Main loop for printing alignment. */
iter = ajListIterNew(hits);
while((hit = (EmbPHit) ajListIterGet(iter)))
{
/* Get pointer to sequence & alignment string. */
ptrp = ajStrGetPtr(hit->Seq);
ptrs = ajStrGetPtr(hit->Alg);
/* Print spacer */
ajFmtPrintF(outf, "# XX\n");
ajFmtPrintF(outf, "# ");
/* if((!sigscanlig_WriteFastaHit(outf, siglist, hits,
hitcnt, ajFalse)))
ajFatal("Bad args to sigscanlig_WriteFasta"); */
if((!sigscanlig_WriteFastaHit(outf, hits, hitcnt, ajFalse)))
ajFatal("Bad args to sigscanlig_WriteFasta");
ajFmtPrintF(outf, "\n# XX\n");
/* Loop for each protein in Hitlist. */
for(num=0, idx=0, y=0;y<niter;y++)
{
num+=mwid2;
ajFmtPrintS(&label, "%S_%d_%d",
hit->Sig->Ligid, hit->Sig->sn, hit->Sig->pn);
/* There is some of the sequence left to print. */
if(idx<MAJSTRGETLEN(hit->Seq))
{
ajFmtPrintF(outf,"%-*S%-*d%-*.*s %d\n",
mwid1, hit->Acc, fwid2,
(num-mwid2+1), mwid2, mwid2, ptrp+idx, num);
ajFmtPrintF(outf,"%-*S%-*c%-*.*s\n",
mwid1, label, fwid2, '-', mwid2,
mwid2, ptrs+idx);
}
/* Printed all the sequence already. */
else
{
ajFmtPrintF(outf,"%-*S%-*d%-*.*s %d\n",
mwid1, hit->Acc, fwid2,
(num-mwid2+1), mwid2, mwid2, ".", num);
ajFmtPrintF(outf,"%-*S%-*c%-*.*s\n",
mwid1, label, fwid2, '-', mwid2,
mwid2, "." );
}
idx += mwid2;
}
hitcnt++;
}
ajListIterDel(&iter);
ajListIterDel(&itersig);
ajStrDel(&label);
return ajTrue;
}
void ajResourceClear(AjPResource resource)
{
AjPReslink lnk = NULL;
AjPResquery qry = NULL;
AjPResterm resterm = NULL;
AjPStr ptr = NULL;
if(MAJSTRGETLEN(resource->Id))
ajStrSetClear(&resource->Id);
if(ajListGetLength(resource->Idalt))
while(ajListstrPop(resource->Idalt,&ptr))
ajStrDel(&ptr);
if(MAJSTRGETLEN(resource->Acc))
ajStrSetClear(&resource->Acc);
if(MAJSTRGETLEN(resource->Name))
ajStrSetClear(&resource->Name);
if(MAJSTRGETLEN(resource->Desc))
ajStrSetClear(&resource->Desc);
if(MAJSTRGETLEN(resource->Url))
ajStrSetClear(&resource->Url);
if(MAJSTRGETLEN(resource->Urllink))
ajStrSetClear(&resource->Urllink);
if(MAJSTRGETLEN(resource->Urlrest))
ajStrSetClear(&resource->Urlrest);
if(MAJSTRGETLEN(resource->Urlsoap))
ajStrSetClear(&resource->Urlsoap);
if(ajListGetLength(resource->Cat))
while(ajListstrPop(resource->Cat,&ptr))
ajStrDel(&ptr);
if(ajListGetLength(resource->Taxon))
while(ajListPop(resource->Taxon,(void**)&resterm))
ajRestermDel(&resterm);
if(ajListGetLength(resource->Edamdat))
while(ajListPop(resource->Edamdat,(void**)&resterm))
ajRestermDel(&resterm);
if(ajListGetLength(resource->Edamfmt))
while(ajListPop(resource->Edamfmt,(void**)&resterm))
ajRestermDel(&resterm);
if(ajListGetLength(resource->Edamid))
while(ajListPop(resource->Edamid,(void**)&resterm))
ajRestermDel(&resterm);
if(ajListGetLength(resource->Edamtpc))
while(ajListPop(resource->Edamtpc,(void**)&resterm))
ajRestermDel(&resterm);
if(ajListGetLength(resource->Xref))
while(ajListPop(resource->Xref,(void**)&lnk))
ajReslinkDel(&lnk);
if(ajListGetLength(resource->Query))
while(ajListPop(resource->Query,(void**)&qry))
ajResqueryDel(&qry);
if(ajListGetLength(resource->Example))
while(ajListstrPop(resource->Example,&ptr))
ajStrDel(&ptr);
if(MAJSTRGETLEN(resource->Db))
ajStrSetClear(&resource->Db);
if(MAJSTRGETLEN(resource->Setdb))
ajStrSetClear(&resource->Setdb);
if(MAJSTRGETLEN(resource->Full))
ajStrSetClear(&resource->Full);
if(MAJSTRGETLEN(resource->Qry))
ajStrSetClear(&resource->Qry);
if(MAJSTRGETLEN(resource->Formatstr))
ajStrSetClear(&resource->Formatstr);
if(MAJSTRGETLEN(resource->Filename))
ajStrSetClear(&resource->Filename);
ajStrDel(&resource->TextPtr);
resource->Count = 0;
resource->Fpos = 0L;
resource->Format = 0;
return;
}
AjPMatrixf ajMatrixfNewFile(const AjPStr filename)
{
AjPMatrixf ret = NULL;
AjPStr *orderstring = NULL;
AjPStr buffer = NULL;
AjPStr firststring = NULL;
AjPStr reststring = NULL;
const AjPStr tok = NULL;
ajint len = 0;
ajint i = 0;
ajint l = 0;
ajint k = 0;
ajint cols = 0;
ajint rows = 0;
const char *ptr = NULL;
AjPFile file = NULL;
AjBool first = ajTrue;
float **matrix = NULL;
float *templine = NULL;
float minval = -1.0;
AjPList rlabel_list = NULL;
AjPStr *rlabel_arr = NULL;
#ifndef WIN32
static const char *delimstr = " :\t\n";
#else
static const char *delimstr = " :\t\n\r";
#endif
rlabel_list = ajListNew();
firststring = ajStrNew();
reststring = ajStrNew();
file = ajDatafileNewInNameS(filename);
if(!file)
{
ajStrDel(&firststring);
ajStrDel(&reststring);
return NULL;
}
/* Read row labels */
while(ajReadline(file,&buffer))
{
ptr = ajStrGetPtr(buffer);
#ifndef WIN32
if(*ptr != '#' && *ptr != '\n')
#else
if(*ptr != '#' && *ptr != '\n' && *ptr != '\r')
#endif
{
if(first)
first = ajFalse;
else
{
ajFmtScanC(ptr, "%S", &firststring);
ajListPushAppend(rlabel_list, firststring);
firststring = ajStrNew();
}
}
}
first = ajTrue;
ajStrDel(&firststring);
rows = ajListToarray(rlabel_list, (void ***) &rlabel_arr);
ajFileSeek(file, 0, 0);
while(ajReadline(file,&buffer))
{
ajStrRemoveWhiteExcess(&buffer);
ptr = ajStrGetPtr(buffer);
if(*ptr && *ptr != '#')
{
if(first)
{
cols = ajStrParseCountC(buffer,delimstr);
AJCNEW0(orderstring, cols);
for(i=0; i<cols; i++)
orderstring[i] = ajStrNew();
tok = ajStrParseC(buffer, " :\t\n");
ajStrAssignS(&orderstring[l++], tok);
while((tok = ajStrParseC(NULL, " :\t\n")))
ajStrAssignS(&orderstring[l++], tok);
first = ajFalse;
ret = ajMatrixfNewAsym(orderstring, cols,
rlabel_arr, rows,
filename);
matrix = ret->Matrixf;
}
else
{
ajFmtScanC(ptr, "%S", &firststring);
/* JISON 19/7/4
k = ajSeqcvtGetCodeK(ret->Cvt,
ajStrGetCharFirst(firststring)); */
k = ajSeqcvtGetCodeS(ret->Cvt, firststring);
len = MAJSTRGETLEN(firststring);
ajStrAssignSubC(&reststring, ptr, len, -1);
/*
** Must discard the first string (label) and use
** reststring otherwise ajArrFloatLine would fail (it
** cannot convert a string to a float)
**
** Use cols,1,cols in below because although 2nd and
** subsequent lines have one more string in them (the
** residue label in the 1st column) we've discarded that
** from the string that's passed
*/
templine = ajArrFloatLine(reststring,delimstr,1,cols);
for(i=0; i<cols; i++)
{
if(templine[i] < minval)
minval = templine[i];
/* JISON 19/7/4
matrix[k][ajSeqcvtGetCodeK(ret->Cvt,
ajStrGetCharFirst(orderstring[i]))]
= templine[i]; */
matrix[k][ajSeqcvtGetCodeAsymS(ret->Cvt,
orderstring[i])]
= templine[i];
}
AJFREE(templine);
}
}
}
ajDebug("fill rest with minimum value %d\n", minval);
ajFileClose(&file);
ajStrDel(&buffer);
for(i=0; i<cols; i++)
ajStrDel(&orderstring[i]);
AJFREE(orderstring);
ajDebug("read matrix file %S\n", filename);
ajStrDel(&firststring);
ajStrDel(&reststring);
for(i=0; i<rows; i++)
ajStrDel(&rlabel_arr[i]);
AJFREE(rlabel_arr);
ajListFree(&rlabel_list);
return ret;
}
int main(ajint argc, char **argv)
{
AjPList pdb_path =NULL; /* Path of pdb files */
AjPStr pdb_name =NULL; /* Name of pdb file */
AjPDirout ccf_path =NULL; /* Path of ccf files */
AjPStr ccf_name =NULL; /* Name of ccf file */
AjPStr pdbid =NULL; /* PDB code */
AjPStr pdbid_temp =NULL; /* PDB code */
AjBool ccfnaming =ajFalse;
/* True == use the pdbid code to name the output file,
False== use the name of the original pdb file*/
/* Mask non-amino acid groups in protein chains that do not
contain a C-alpha atom. The group will not appear in either
the CO or SQ records of the clean coordinate file */
AjBool camask =ajFalse;
/* Mask amino acids in protein chains that do not contain a
C-alpha atom. The amino acid will appear not appear in the
CO record but will still be present in the SQ record of the
clean coordinate file */
AjBool camask1 =ajFalse;
/* Mask residues or groups in protein chains with a single atom only */
AjBool atommask =ajFalse;
AjPStr temp =NULL; /* Temp string */
AjPStr msg =NULL; /* Error message */
AjPStr base_name =NULL; /* Name of pdb file w/o path or
extension */
AjPFile pdb_inf =NULL; /* pdb input file pointer */
AjPFile ccf_outf =NULL; /* ccf output file pointer */
AjPFile logf =NULL; /* log file pointer*/
AjPPdb pdb =NULL; /* Pdb structure (for parsed data) */
ajint min_chain_size=0; /* Minimum length of a SEQRES chain
for it to be parsed */
ajint max_mismatch=0; /* Max. no. residues to trim when checking
for missing C-terminal SEQRES residues. */
ajint max_trim=0; /* Maximum number of permissible
mismatches between the ATOM and
SEQRES sequences */
ajint pos =0; /* Location of the file extension in
the pdb file name */
/* THIS_DIAGNOSTIC
tempstr=ajStrNew();
ajStrAssignC(&tempstr, "diagnostics");
tempfile=ajFileNewOutNameS(tempstr);
ajStrDel(&tempstr);*/
/* Initialise strings */
ccf_name = ajStrNew();
pdb_name = ajStrNew();
temp = ajStrNew();
msg = ajStrNew();
base_name = ajStrNew();
pdbid = ajStrNew();
pdbid_temp = ajStrNew();
/* Read data from acd */
embInitPV("pdbparse",argc,argv,"STRUCTURE",VERSION);
pdb_path = ajAcdGetDirlist("pdbpath");
ccf_path = ajAcdGetOutdir("ccfoutdir");
logf = ajAcdGetOutfile("logfile");
min_chain_size=ajAcdGetInt("chnsiz");
max_mismatch =ajAcdGetInt("maxmis");
max_trim =ajAcdGetInt("maxtrim");
ccfnaming = ajAcdGetBoolean("ccfnaming");
camask = ajAcdGetBoolean("camask");
camask1 = ajAcdGetBoolean("camaska");
atommask = ajAcdGetBoolean("atommask");
/* Check directories*/
/*Start of main application loop*/
while(ajListPop(pdb_path,(void **)&temp))
{
ajFmtPrint("Processing %S\n", temp);
ajFmtPrintF(logf, "%S\n", temp);
/* Read pdb file*/
if((pdb_inf=ajFileNewInNameS(temp))==NULL)
{
ajFmtPrintS(&msg, "Could not open for reading %S ",
temp);
ajWarn(ajStrGetPtr(msg));
ajFmtPrintF(logf, "%-15s%S\n//\n", "FILE_OPEN", temp);
ajStrDel(&temp);
continue;
}
/* Assign pdb id code from file name */
ajStrAssignS(&pdbid, temp);
ajFilenameTrimPathExt(&pdbid);
if(MAJSTRGETLEN(pdbid)>4)
{
/* The file name is longer than expected (and probably contains a
prefix). Take the last four characters to be the pdbid code */
ajStrAssignSubS(&pdbid_temp, pdbid, pos-4, pos-1);
ajStrAssignS(&pdbid, pdbid_temp);
}
else if(MAJSTRGETLEN(pdbid)<4)
ajFatal("Could not determine pdbid code from file name (%S)", pdbid);
/* Parse pdb file and write pdb structure */
if(!(pdb=ajPdbReadRawNew(pdb_inf, pdbid, min_chain_size, max_mismatch,
max_trim, camask, camask1, atommask, logf)))
{
ajFmtPrintS(&msg, "Clean coordinate file not generated for %S", temp);
ajWarn(ajStrGetPtr(msg));
ajFmtPrintF(logf, "%-15s%S\n//\n", "NO_OUTPUT", temp);
ajFileClose(&pdb_inf);
ajStrDel(&temp);
continue;
}
/* Open clean coordinate file for writing*/
if(ccfnaming)
ajStrAssignS(&ccf_name, pdb->Pdb);
else
ajStrAssignS(&ccf_name, temp);
ajStrFmtLower(&ccf_name);
if(!(ccf_outf=ajFileNewOutNameDirS(ccf_name, ccf_path)))
{
ajFmtPrintS(&msg, "Could not open %S for writing",
ccf_name);
ajWarn(ajStrGetPtr(msg));
ajFmtPrintF(logf, "%-15s%S\n//\n", "FILE_OPEN", ccf_name);
ajFileClose(&pdb_inf);
ajPdbDel(&pdb);
ajStrDel(&temp);
continue;
}
/* Write pdb file */
if(!ajPdbWriteAll(ccf_outf, pdb))
{
ajFmtPrintS(&msg, "Could not write file %S", ccf_name);
ajWarn(ajStrGetPtr(msg));
ajFmtPrintF(logf, "%-15s%S\n//\n", "FILE_WRITE", ccf_name);
ajFmtPrintS(&temp, "rm %S", ccf_name);
ajFmtPrint("%S", temp);
ajSysSystem(temp);
ajFileClose(&pdb_inf);
ajFileClose(&ccf_outf);
ajPdbDel(&pdb);
ajStrDel(&temp);
continue;
}
/* Tidy up*/
ajFileClose(&pdb_inf);
ajFileClose(&ccf_outf);
ajPdbDel(&pdb);
ajStrDel(&temp);
ajFmtPrintF(logf, "//\n");
}
/*End of main application loop*/
/*Tidy up */
ajListFree(&pdb_path);
ajStrDel(&pdb_name);
ajDiroutDel(&ccf_path);
ajStrDel(&ccf_name);
ajStrDel(&base_name);
ajStrDel(&pdbid);
ajStrDel(&pdbid_temp);
ajStrDel(&msg);
ajFileClose(&logf);
/* DIAGNOSTIC
ajFileClose(&tempfile);
*/
/* Return */
ajExit();
return 0;
}
int main(int argc, char **argv)
{
/* ACD data item variables */
AjPSeqset dataset = NULL;
AjPFile bfile = NULL;
AjPFile plib = NULL;
AjPStr mod = NULL;
ajint nmotifs = 0;
AjBool text = ajFalse;
AjPStr prior = NULL;
float evt = 0.0;
ajint nsites = 0;
ajint minsites = 0;
ajint maxsites = 0;
float wnsites = 0.0;
ajint w = 0;
ajint minw = 0;
ajint maxw = 0;
AjBool nomatrim = ajFalse;
ajint wg = 0;
ajint ws = 0;
AjBool noendgaps = ajFalse;
AjBool revcomp = ajFalse;
AjBool pal = ajFalse;
AjBool nostatus = ajFalse;
ajint maxiter = 0;
float distance = 0.0;
float b = 0.0;
float spfuzz = 0.0;
AjPStr spmap = NULL;
AjPStr cons = NULL;
ajint maxsize = 0;
ajint p = 0;
ajint time = 0;
AjPStr sf = NULL;
ajint heapsize = 64;
AjBool xbranch = ajFalse;
AjBool wbranch = ajFalse;
ajint bfactor = 0;
AjPFile outtext = NULL;
/* Housekeeping variables */
AjPStr cmd = NULL;
AjPStr ssname = NULL;
AjPSeqout outseq = NULL;
AjPStr tmp = NULL;
char option;
/* ACD file processing */
embInitPV("ememetext",argc,argv,"MEME",VERSION);
dataset = ajAcdGetSeqset("dataset");
bfile = ajAcdGetInfile("bfile");
plib = ajAcdGetInfile("plibfile");
mod = ajAcdGetSelectSingle("mod");
nmotifs = ajAcdGetInt("nmotifs");
text = ajAcdGetBoolean("text");
prior = ajAcdGetSelectSingle("prior");
evt = ajAcdGetFloat("evt");
nsites = ajAcdGetInt("nsites");
minsites = ajAcdGetInt("minsites");
maxsites = ajAcdGetInt("maxsites");
wnsites = ajAcdGetFloat("wnsites");
w = ajAcdGetInt("w");
minw = ajAcdGetInt("minw");
maxw = ajAcdGetInt("maxw");
nomatrim = ajAcdGetBoolean("nomatrim");
wg = ajAcdGetInt("wg");
ws = ajAcdGetInt("ws");
noendgaps = ajAcdGetBoolean("noendgaps");
revcomp = ajAcdGetBoolean("revcomp");
pal = ajAcdGetBoolean("pal");
nostatus = ajAcdGetBoolean("nostatus");
maxiter = ajAcdGetInt("maxiter");
distance = ajAcdGetFloat("distance");
b = ajAcdGetFloat("b");
spfuzz = ajAcdGetFloat("spfuzz");
spmap = ajAcdGetSelectSingle("spmap");
cons = ajAcdGetString("cons");
maxsize = ajAcdGetInt("maxsize");
p = ajAcdGetInt("p");
time = ajAcdGetInt("time");
sf = ajAcdGetString("sf");
heapsize = ajAcdGetInt("heapsize");
xbranch = ajAcdGetBoolean("xbranch");
wbranch = ajAcdGetBoolean("wbranch");
bfactor = ajAcdGetInt("bfactor");
outtext = ajAcdGetOutfile("outtext");
outseq = ajAcdGetSeqoutset("outseq");
/* MAIN APPLICATION CODE */
/* 1. Housekeeping */
cmd = ajStrNew();
tmp = ajStrNew();
/* 2. Re-write dataset to a temporary file in a format (fasta) MEME
** can understand.
** Can't just pass the name of dataset to MEME as the name provided
** might be a USA which MEME would not understand.
*/
ssname = ajStrNewS(ajFileGetNameS(outseq->File));
ajSeqoutSetFormatC(outseq, "fasta");
ajSeqoutWriteSet(outseq, dataset);
ajSeqoutClose(outseq);
ajSeqoutDel(&outseq);
/* 3. Build ememe command line */
/* Command line is built in this order:
i. Application name.
ii. Original MEME options (in order they appear in ACD file)
iii.Original MEME options (that don't appear in ACD file)
iv. EMBASSY MEME new qualifiers and parameters.
*/
ajStrAssignS(&cmd, ajAcdGetpathC("meme"));
ajFmtPrintAppS(&cmd, " %S", ssname);
if(bfile)
ajFmtPrintAppS(&cmd, " -bfile %s ", ajFileGetNameC(bfile));
if(plib)
ajFmtPrintAppS(&cmd, " -plib %s ", ajFileGetNameC(plib));
option = ajStrGetCharFirst(mod);
if(option == 'o')
ajStrAppendC(&cmd, " -mod oops ");
else if(option == 'z')
ajStrAppendC(&cmd, " -mod zoops ");
else if(option == 'a')
ajStrAppendC(&cmd, " -mod anr ");
if(nmotifs != 1)
ajFmtPrintAppS(&cmd, " -nmotifs %d ", nmotifs);
if(text)
ajFmtPrintAppS(&cmd, " -text ");
ajFmtPrintAppS(&cmd, " -prior %S ", prior);
if(evt != -1)
ajFmtPrintAppS(&cmd, " -evt %f ", evt);
if(nsites != -1)
ajFmtPrintAppS(&cmd, " -nsites %d ", nsites);
else
{
if(minsites != -1)
ajFmtPrintAppS(&cmd, " -minsites %d ", minsites);
if(maxsites != -1)
ajFmtPrintAppS(&cmd, " -maxsites %d ", maxsites);
}
if(wnsites < 0.7999 || wnsites > .8001)
ajFmtPrintAppS(&cmd, " -wnsites %f ", wnsites);
if(w != -1)
ajFmtPrintAppS(&cmd, " -w %d ", w);
if(minw != 8)
ajFmtPrintAppS(&cmd, " -minw %d ", minw);
if(maxw != 50)
ajFmtPrintAppS(&cmd, " -maxw %d ", maxw);
if(nomatrim)
ajFmtPrintAppS(&cmd, " -nomatrim ");
if(wg != 11)
ajFmtPrintAppS(&cmd, " -wg %d ", wg);
if(ws != 1)
ajFmtPrintAppS(&cmd, " -ws %d ", ws);
if(noendgaps)
ajFmtPrintAppS(&cmd, " -noendgaps ");
if(revcomp)
ajFmtPrintAppS(&cmd, " -revcomp ");
if(pal && ajSeqsetIsNuc(dataset))
ajFmtPrintAppS(&cmd, " -pal ");
if(nostatus)
ajFmtPrintAppS(&cmd, " -nostatus ");
if(maxiter != 50)
ajFmtPrintAppS(&cmd, " -maxiter %d ", maxiter);
if(distance < 0.00099 || distance > 0.00101)
ajFmtPrintAppS(&cmd, " -distance %f ", distance);
if(b != -1)
ajFmtPrintAppS(&cmd, " -b %f ", b);
if(spfuzz != -1)
ajFmtPrintAppS(&cmd, " -spfuzz %f ", spfuzz);
if(!ajStrMatchC(spmap,"default"))
ajFmtPrintAppS(&cmd, " -spmap %S ", spmap);
if(MAJSTRGETLEN(cons))
ajFmtPrintAppS(&cmd, "-cons %S", cons);
if(maxsize != -1)
ajFmtPrintAppS(&cmd, " -maxsize %d ", maxsize);
if(p > 0)
ajFmtPrintAppS(&cmd, " -p %d ", p);
if(time > 0)
ajFmtPrintAppS(&cmd, " -time %d ", time);
if(MAJSTRGETLEN(sf))
ajFmtPrintAppS(&cmd, " -sf %S", sf);
if(heapsize != 64)
ajFmtPrintAppS(&cmd, " -heapsize %d ", heapsize);
if(xbranch)
ajFmtPrintAppS(&cmd, " -x_branch");
if(wbranch)
ajFmtPrintAppS(&cmd, " -w_branch");
if(bfactor != 3)
ajFmtPrintAppS(&cmd, " -bfactor %d ", bfactor);
if(ajSeqsetIsProt(dataset))
ajFmtPrintAppS(&cmd, "-protein ");
else
ajFmtPrintAppS(&cmd, "-dna ");
ajFmtPrintAppS(&cmd, " -text");
ajFmtPrintAppS(&cmd, " > %S ", ajFileGetNameS(outtext));
/* 4. Close files from ACD before calling meme */
ajFileClose(&bfile);
ajFileClose(&plib);
/* 5. Call meme */
/* ajFmtPrint("\n%S\n", cmd); */
system(ajStrGetPtr(cmd));
/* 6. Exit cleanly */
ajSeqsetDel(&dataset);
ajStrDel(&cons);
ajStrDel(&sf);
ajStrDel(&mod);
ajStrDel(&prior);
ajStrDel(&spmap);
ajStrDel(&cmd);
ajStrDel(&ssname);
ajStrDel(&tmp);
ajFileClose(&bfile);
ajFileClose(&plib);
ajFileClose(&outtext);
ajSeqoutDel(&outseq);
embExit();
return 0;
}
/* @funcstatic domainalign_keepsinglets ***************************************
**
** Write singlet sequences to file.
**
** @param [r] domain [AjPDomain]
** @param [r] noden [ajint]
** @param [r] singlets [AjPDirout]
** @param [r] logf [AjPFile]
**
** @return [void] True on success
** @@
****************************************************************************/
static void domainalign_keepsinglets(AjPDomain domain,
ajint noden,
AjPDirout singlets,
AjPFile logf)
{
AjPStr temp2 = NULL; /* A temporary string. */
AjPFile singf = NULL; /* File pointer for singlets file. */
EmbPHitlist hitlist = NULL; /* Hitlist object for output of data. */
temp2 = ajStrNew();
domainalign_writesid(domain, noden, &temp2);
if(MAJSTRGETLEN(ajDomainGetSeqPdb(domain)))
{
/* Write Hit object. */
hitlist = embHitlistNew(1);
hitlist->Type = domain->Type;
ajStrAssignS(&hitlist->hits[0]->Seq, ajDomainGetSeqPdb(domain));
ajStrAssignS(&hitlist->hits[0]->Acc, ajDomainGetAcc(domain));
ajStrAssignS(&hitlist->hits[0]->Spr, ajDomainGetSpr(domain));
ajStrAssignS(&hitlist->hits[0]->Dom, ajDomainGetId(domain));
if((domain->Type == ajSCOP))
{
ajStrAssignS(&hitlist->Class, domain->Scop->Class);
ajStrAssignS(&hitlist->Fold, domain->Scop->Fold);
ajStrAssignS(&hitlist->Superfamily, domain->Scop->Superfamily);
ajStrAssignS(&hitlist->Family, domain->Scop->Family);
if(noden==4)
hitlist->Sunid_Family = domain->Scop->Sunid_Family;
else if(noden==3)
hitlist->Sunid_Family = domain->Scop->Sunid_Superfamily;
else if(noden==2)
hitlist->Sunid_Family = domain->Scop->Sunid_Fold;
else if(noden==1)
hitlist->Sunid_Family = domain->Scop->Sunid_Class;
}
else
{
ajStrAssignS(&hitlist->Class, domain->Cath->Class);
ajStrAssignS(&hitlist->Architecture, domain->Cath->Architecture);
ajStrAssignS(&hitlist->Topology, domain->Cath->Topology);
ajStrAssignS(&hitlist->Superfamily, domain->Cath->Superfamily);
if(noden==9)
hitlist->Sunid_Family = domain->Cath->Family_Id;
if(noden==8)
hitlist->Sunid_Family = domain->Cath->Superfamily_Id;
else if(noden==7)
hitlist->Sunid_Family = domain->Cath->Topology_Id;
else if(noden==6)
hitlist->Sunid_Family = domain->Cath->Arch_Id;
else if(noden==5)
hitlist->Sunid_Family = domain->Cath->Class_Id;
}
singf = ajFileNewOutNameDirS(temp2, singlets);
embHitlistWriteFasta(singf, hitlist);
/* ajFmtPrintF(singf, ">%S\n%S\n", temp2, ajDomainGetSeqPdb(domain)); */
ajFileClose(&singf);
}
else
{
ajWarn("No sequence found (no 'DS' record in DCF file) for singlet %S "
"despite configuring to save singlets. "
"NO file written!", ajDomainGetId(domain));
ajFmtPrintF(logf, "No sequence found (no 'DS' record in DCF file) for "
"singlet %S despite configuring to save singlets. "
"NO file written!", ajDomainGetId(domain));
}
ajStrDel(&temp2);
return;
}
static AjBool dbxflat_ParseFastq(EmbPBtreeEntry entry, AjPFile inf)
{
AjPStr line = NULL;
ajlong pos = 0L;
ajuint seqlen = 0;
ajuint qlen = 0;
AjPStr tmpfd = NULL;
AjPStr str = NULL;
AjPStr de = NULL;
AjBool ok;
if(!dbxflat_wrdexp)
dbxflat_wrdexp = ajRegCompC("([A-Za-z0-9.:=]+)");
line = ajStrNewC("");
pos = ajFileResetPos(inf);
if(!ajReadlineTrim(inf,&line))
{
ajStrDel(&line);
return ajFalse;
}
/* first line of entry */
if(!ajStrPrefixC(line,"@"))
return ajFalse;
entry->fpos = pos;
ajStrCutStart(&line, 1);
ajStrExtractFirst(line, &de, &entry->id);
if(desfield && ajStrGetLen(de))
{
while(ajRegExec(dbxflat_wrdexp,de))
{
ajRegSubI(dbxflat_wrdexp, 1, &tmpfd);
str = ajStrNew();
ajStrAssignS(&str,tmpfd);
ajListPush(desfield->data,(void *)str);
ajRegPost(dbxflat_wrdexp, &de);
}
}
/* now read sequence */
ok = ajReadlineTrim(inf,&line);
while(ok && !ajStrPrefixC(line, "+"))
{
ajStrRemoveWhite(&line);
seqlen += MAJSTRGETLEN(line);
ok = ajReadlineTrim(inf,&line);
}
if(!ok)
return ajFalse;
ok = ajReadlineTrim(inf,&line);
while(ok)
{
qlen += MAJSTRGETLEN(line);
if(qlen < seqlen)
ok = ajReadlineTrim(inf,&line);
else
ok = ajFalse;
}
ajStrDel(&de);
ajStrDel(&tmpfd);
ajStrDel(&line);
return ajTrue;
}
int main(int argc, char **argv)
{
/* Variable declarations */
AjPStr dbname = NULL;
AjPStr svrname = NULL;
AjPFile outfile = NULL;
AjPResource resource = NULL;
AjPResourcein resourcein = NULL;
AjPStr resourceqry = NULL;
AjPStr type = NULL;
AjBool id;
AjBool qry;
AjBool all;
AjBool verbose;
AjPStr methods = NULL;
AjPStr release = NULL;
AjPStr comment = NULL;
AjPStr defined = NULL;
AjPList list = NULL;
AjPTagval tagval = NULL;
AjIList iter = NULL;
ajuint space = 0;
AjPList aliaslist = NULL;
AjIList aliter = NULL;
const AjPStr alias = NULL;
ajuint maxlen;
AjPStr truedbname = NULL;
AjBool isalias = ajFalse;
/* ACD processing */
embInit("dbtell", argc, argv);
dbname = ajAcdGetString("database");
svrname = ajAcdGetString("server");
verbose = ajAcdGetBoolean("full");
outfile = ajAcdGetOutfile("outfile");
ajStrAssignS(&truedbname, dbname);
ajNamAliasDatabase(&truedbname);
ajStrFmtLower(&truedbname);
if(!ajStrMatchS(dbname, truedbname))
isalias = ajTrue;
/* Application logic */
/* Check EMBOSS database information.
Write output file */
if(ajNamDbDetailsSvr(truedbname, svrname, &type, &id, &qry, &all,
&comment, &release, &methods, &defined))
{
if(isalias)
ajFmtPrintF(outfile, "# %S is an alias for %S defined in %S\n",
dbname, truedbname, defined);
else
ajFmtPrintF(outfile, "# %S is defined in %S\n",
truedbname, defined);
ajFmtPrintF(outfile, "# access levels id: %B query: %B all: %B\n\n",
id, qry, all);
ajFmtPrintF(outfile, "DBNAME %S [\n", truedbname);
if(ajStrGetLen(svrname))
list = ajNamDbGetAttrlistSvr(truedbname, svrname);
else
list = ajNamDbGetAttrlist(truedbname);
iter = ajListIterNewread(list);
while(!ajListIterDone(iter))
{
tagval = ajListIterGet(iter);
space = 15 - ajStrGetLen(ajTagvalGetTag(tagval));
ajFmtPrintF(outfile, " %S:%.*s\"%S\"\n",
ajTagvalGetTag(tagval),
space, " ",
ajTagvalGetValue(tagval));
ajTagvalDel(&tagval);
}
ajListIterDel(&iter);
ajListFree(&list);
ajFmtPrintF(outfile, "]\n");
if(verbose)
{
aliaslist = ajListNew();
ajNamListFindAliases(truedbname, aliaslist);
if(ajListGetLength(aliaslist))
{
ajFmtPrintF(outfile, "\n");
aliter = ajListIterNewread(aliaslist);
maxlen = 1;
while(!ajListIterDone(aliter))
{
alias = ajListIterGet(aliter);
if(MAJSTRGETLEN(alias) > maxlen)
maxlen = MAJSTRGETLEN(alias);
}
ajListIterDel(&aliter);
aliter = ajListIterNewread(aliaslist);
while(!ajListIterDone(aliter))
{
alias = ajListIterGet(aliter);
if(ajStrFindK(alias, ':') < 0)
ajFmtPrintF(outfile, "ALIAS %-*S %S\n",
maxlen, alias, truedbname);
}
ajListIterDel(&aliter);
}
ajListstrFree(&aliaslist);
}
ajStrDel(&type);
ajStrDel(&methods);
ajStrDel(&release);
ajStrDel(&comment);
ajStrDel(&defined);
}
else
{
/* try looking in DRCAT */
resourcein = ajResourceinNew();
resource = ajResourceNew();
ajFmtPrintS(&resourceqry, "drcat:%S", dbname);
ajResourceinQryS(resourcein, resourceqry);
if(ajResourceinRead(resourcein, resource))
{
ajFmtPrintF(outfile, "DBNAME %S [\n", dbname);
ajFmtPrintF(outfile, " comment: \"defined in DRCAT\"\n");
ajFmtPrintF(outfile, "]\n");
}
ajResourceinDel(&resourcein);
ajResourceDel(&resource);
ajStrDel(&resourceqry);
}
/* Memory clean-up and exit */
ajFileClose(&outfile);
ajStrDel(&truedbname);
ajStrDel(&dbname);
ajStrDel(&svrname);
embExit();
return 0;
}
