AjPMatrix ajMatrixNew(const AjPPStr codes, ajint n, const AjPStr filename)
{
ajint i = 0;
AjPMatrix ret = NULL;
ajint nsize;
if((!n) || (!codes) || (!filename))
return NULL;
nsize = n + 1;
AJNEW0(ret);
ajStrAssignS(&ret->Name, filename);
AJCNEW0(ret->Codes, n);
for(i=0; i<n; i++)
ret->Codes[i] = ajStrNew();
for(i=0; i<n; i++)
ajStrAssignS(&ret->Codes[i], codes[i]);
ret->Size = nsize;
AJCNEW0(ret->Matrix, nsize);
for(i=0; i<nsize; i++)
AJCNEW0(ret->Matrix[i], nsize);
ret->Cvt = ajSeqcvtNewStr(codes, n);
return ret;
}
AjPMatrixf ajMatrixfNewAsym(const AjPPStr codes, ajint n,
const AjPPStr rcodes, ajint rn,
const AjPStr filename)
{
ajint i = 0;
AjPMatrixf ret = 0;
ajint nsize;
ajint rnsize;
if((!n) || (!codes) || (!filename))
return NULL;
nsize = n + 1;
rnsize = rn + 1;
AJNEW0(ret);
ajStrAssignS(&ret->Name, filename);
AJCNEW0(ret->Codes, n);
for(i=0; i<n; i++)
ret->Codes[i] = ajStrNew();
for(i=0; i<n; i++)
ajStrAssignS(&ret->Codes[i], codes[i]);
ret->Size = nsize;
AJCNEW0(ret->CodesRow, rn);
for(i=0; i<rn; i++)
ret->CodesRow[i] = ajStrNew();
for(i=0; i<rn; i++)
ajStrAssignS(&ret->CodesRow[i], rcodes[i]);
ret->SizeRow = rnsize;
AJCNEW0(ret->Matrixf, rnsize);
for(i=0; i<rnsize; i++)
AJCNEW0(ret->Matrixf[i], nsize);
ret->Cvt = ajSeqcvtNewStrAsym(codes, n, rcodes, rn);
return ret;
}
AjPRegexp ajRegCompC(const char* rexp)
{
AjPRegexp ret;
int options = 0;
int errpos = 0;
const char *errptr = NULL;
const unsigned char *tableptr = NULL;
AJNEW0(ret);
AJCNEW0(ret->ovector, AJREG_OVECSIZE);
ret->ovecsize = AJREG_OVECSIZE/3;
ret->pcre = pcre_compile(rexp, options, &errptr, &errpos, tableptr);
if(!ret->pcre)
{
ajErr("Failed to compile regular expression '%s' at position %d: %s",
rexp, errpos, errptr);
AJFREE(ret);
return NULL;
}
regAlloc += sizeof(ret);
regCount ++;
regTotal ++;
/*ajDebug("ajRegCompC %x size %d regexp '%s'\n",
ret, (int) sizeof(ret), rexp);*/
return ret;
}
static ajuint jaspscan_readmatrix(const AjPStr mfname, float ***matrix)
{
AjPFile inf = NULL;
AjPStr line = NULL;
ajuint i = 0;
ajuint cols = 0;
AJCNEW0(*matrix,4);
line = ajStrNew();
inf = ajFileNewInNameS(mfname);
if(!inf)
ajFatal("Cannot open matrix file %S",mfname);
i = 0;
while(ajReadlineTrim(inf,&line))
{
if(!i)
cols = ajStrParseCountC(line," \n");
(*matrix)[i++] = ajArrFloatLine(line," \n",1,cols);
}
ajStrDel(&line);
ajFileClose(&inf);
return cols;;
}
AjPHist ajHistNew(ajuint numofsets, ajuint numofpoints)
{
static AjPHist hist = NULL;
ajuint i;
AJNEW0(hist);
hist->numofsets = 0;
hist->numofsetsmax = numofsets;
hist->numofdatapoints = numofpoints;
hist->xmin = 0;
hist->xmax = 0;
hist->displaytype = HIST_SIDEBYSIDE; /* default draw
multiple histograms
side by side */
hist->bins = 0;
hist->BaW = AJFALSE;
ajStrAssignEmptyC(&hist->title,"");
ajStrAssignEmptyC(&hist->xaxis,"");
ajStrAssignEmptyC(&hist->yaxisleft,"");
ajStrAssignEmptyC(&hist->yaxisright,"");
AJCNEW0(hist->hists,numofsets);
for(i=0;i<numofsets; i++)
{
AJNEW0((hist->hists[i]));
(hist->hists)[i]->data = NULL;
(hist->hists)[i]->deletedata = AJFALSE;
(hist->hists)[i]->colour = i+2;
(hist->hists)[i]->pattern = 0;
(hist->hists)[i]->title = NULL;
(hist->hists)[i]->xaxis = NULL;
(hist->hists)[i]->yaxis = NULL;
}
return hist;
}
EmbPPropAmino* embPropEaminoRead(AjPFile mfptr)
{
AjPStr line = NULL;
AjPStr token = NULL;
AjBool firstline;
const char *p;
ajuint i;
ajint n;
EmbPPropAmino *ret;
line = ajStrNew();
token = ajStrNew();
firstline = ajTrue;
AJCNEW0(ret,EMBPROPSIZE);
for(i=0; i < EMBPROPSIZE; ++i)
AJNEW0(ret[i]);
while(ajReadline(mfptr, &line))
{
ajStrRemoveWhiteExcess(&line);
p = ajStrGetPtr(line);
if(*p=='#' || *p=='!' || !*p)
continue;
if(firstline)
{
if(!ajStrPrefixC(line,"aa"))
ajFatal("Incorrect (old?) format amino data file");
firstline = ajFalse;
continue;
}
ajFmtScanS(line,"%S",&token);
ajStrFmtUpper(&token);
if(ajStrGetLen(token) != 1)
ajFatal("Amino file line doesn't begin with a single character");
i = ajBasecodeToInt((ajint) *ajStrGetPtr(token));
if(i == 27)
ajFatal("Amino file line doesn't begin with a single A->Z (%S)",
line);
n = ajFmtScanS(line,"%*s%d%d%d%d%d%d%f%d%d%d",
&ret[i]->tiny,
&ret[i]->sm_all,
&ret[i]->aliphatic,
&ret[i]->aromatic,
&ret[i]->nonpolar,
&ret[i]->polar,
&ret[i]->charge,
&ret[i]->pve,
&ret[i]->nve,
&ret[i]->extcoeff);
if(n!= 10)
ajFatal("Only %d columns in amino file - expected %d",n+1,11);
}
ajStrDel(&line);
ajStrDel(&token);
return ret;
}
static AjBool assemoutWriteNextBam(AjPOutfile outfile, const AjPAssem assem)
{
AjPFile outf = ajOutfileGetFile(outfile);
AjPSeqBamHeader header = NULL;
AjPAssemContig c = NULL;
AjPSeqBam bam;
AjPAssemRead r = NULL;
AjPAssemContig* contigs = NULL;
AjPAssemTag t = NULL;
AjIList j = NULL;
AjPSeqBamBgzf gzfile = NULL;
AjPStr headertext=NULL;
const AjPStr rgheadertext=NULL;
AjBool ret = ajTrue;
ajint i=0;
ajulong ncontigs=0UL;
if(!outf) return ajFalse;
if(!assem) return ajFalse;
if(!assem->Hasdata)
{
if(ajListGetLength(assem->ContigsOrder))
ncontigs = ajListToarray(assem->ContigsOrder, (void***)&contigs);
else
ncontigs = ajTableToarrayValues(assem->Contigs, (void***)&contigs);
ajFmtPrintS(&headertext, "@HD\tVN:1.3\tSO:%s\n",
ajAssemGetSortorderC(assem));
header = ajSeqBamHeaderNewN((ajuint) ncontigs);
gzfile = ajSeqBamBgzfNew(ajFileGetFileptr(outf), "w");
outfile->OutData = gzfile;
while (contigs[i]) /* contigs */
{
c = contigs[i];
if(ajStrMatchC(c->Name, "*"))
{
i++;
continue;
}
header->target_name[i] = strdup(ajStrGetPtr(c->Name));
header->target_len[i++] = c->Length;
ajFmtPrintAppS(&headertext, "@SQ\tSN:%S\tLN:%d",
c->Name, c->Length);
if(c->URI)
ajFmtPrintAppS(&headertext, "\tUR:%S", c->URI);
if(c->MD5)
ajFmtPrintAppS(&headertext, "\tM5:%S", c->MD5);
if(c->Species)
ajFmtPrintAppS(&headertext, "\tSP:%S", c->Species);
ajFmtPrintAppS(&headertext, "\n");
j = ajListIterNewread(c->Tags);
while (!ajListIterDone(j))
{
t = ajListIterGet(j);
ajFmtPrintAppS(&headertext,
"@CO\t%S %u %u %S\n", t->Name, t->x1, t->y1,
t->Comment);
}
ajListIterDel(&j);
}
rgheadertext = assemSAMGetReadgroupHeaderlines(assem);
if(rgheadertext)
ajStrAppendS(&headertext, rgheadertext);
ajSeqBamHeaderSetTextC(header, ajStrGetPtr(headertext));
ajSeqBamHeaderWrite(gzfile, header);
ajSeqBamHeaderDel(&header);
ajStrDel(&headertext);
AJFREE(contigs);
if(!assem->BamHeader)
return ajTrue;
}
/* data */
gzfile = outfile->OutData;
AJNEW0(bam);
bam->m_data=10;
AJCNEW0(bam->data, bam->m_data);
j = ajListIterNewread(assem->Reads);
while (!ajListIterDone(j)) /* reads */
{
r = ajListIterGet(j);
assemoutWriteBamAlignment(gzfile, r, bam);
}
ajListIterDel(&j);
AJFREE(bam->data);
AJFREE(bam);
/* ajSeqBamBgzfClose(gzfile);*/
return ret;
}
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(int argc, char **argv)
{
AjPList idlist;
AjPList* fieldList = NULL;
AjBool systemsort;
AjBool cleanup;
ajint blastv = 0;
char dbtype = '\0';
ajuint maxindex;
ajuint maxidlen = 0;
ajuint maxlen;
AjPStr version = NULL;
AjPStr seqtype = NULL;
AjPFile elistfile = NULL;
AjPFile* alistfile = NULL;
AjPStr dbname = NULL;
AjPStr release = NULL;
AjPStr datestr = NULL;
AjPStr sortopt = NULL;
void **entryIds = NULL;
AjBool usesrc = AJTRUE;
AjPStr directory;
AjPStr indexdir;
AjPStr filename;
AjPStr exclude;
AjPStr curfilename = NULL;
AjPStr idformat = NULL;
EmbPEntry entry;
PBlastDb db = NULL;
ajuint idCount = 0;
ajuint idDone;
AjPList listTestFiles = NULL;
void ** testFiles = NULL;
ajuint nfiles;
ajuint ifile;
ajuint jfile;
ajuint filesize;
short recsize;
ajuint maxfilelen = 20;
char date[4] =
{
0,0,0,0
};
AjPStr tmpfname = NULL;
AjPStr* fields = NULL;
AjPFile entFile = NULL;
AjPStr* divfiles = NULL;
ajint* maxFieldLen = NULL;
ajuint ifield = 0;
ajuint nfields = 0;
AjPFile logfile = NULL;
ajuint* countField = NULL;
ajuint* fieldTot = NULL;
ajuint idCountFile = 0;
ajuint i = 0;
embInit("dbiblast", argc, argv);
idformat = ajStrNewC("NCBI");
fields = ajAcdGetList("fields");
directory = ajAcdGetDirectoryName("directory");
indexdir = ajAcdGetOutdirName("indexoutdir");
filename = ajAcdGetString("filenames");
exclude = ajAcdGetString("exclude");
dbname = ajAcdGetString("dbname");
release = ajAcdGetString("release");
datestr = ajAcdGetString("date");
systemsort = ajAcdGetBoolean("systemsort");
cleanup = ajAcdGetBoolean("cleanup");
sortopt = ajAcdGetString("sortoptions");
maxindex = ajAcdGetInt("maxindex");
version = ajAcdGetListSingle("blastversion");
seqtype = ajAcdGetListSingle("seqtype");
usesrc = ajAcdGetBoolean("sourcefile");
logfile = ajAcdGetOutfile("outfile");
while(fields[nfields]) /* array ends with a NULL */
nfields++;
if(nfields)
{
AJCNEW(maxFieldLen, nfields);
AJCNEW0(countField, nfields);
AJCNEW0(fieldTot, nfields);
for(ifield=0; ifield < nfields; ifield++)
maxFieldLen[ifield] = (ajint) maxindex * -1;
if(systemsort)
AJCNEW(alistfile, nfields);
else
{
AJCNEW(fieldList, nfields);
for(ifield=0; ifield < nfields; ifield++)
fieldList[ifield] = ajListNew();
}
}
if(ajStrMatchC(datestr, "00/00/00"))
ajFmtPrintS(&datestr, "%D", ajTimeRefTodayFmt("dbindex"));
ajStrRemoveWhite(&dbname); /* used for temp filenames */
embDbiDateSet(datestr, date);
idlist = ajListNew();
if(ajUtilGetBigendian())
readReverse = ajFalse;
else
readReverse = ajTrue;
ajStrToInt(version, &blastv);
dbtype = ajStrGetCharFirst(seqtype);
ajDebug("reading '%S/%S'\n", directory, filename);
ajDebug("writing '%S/'\n", indexdir);
listTestFiles = embDbiFileListExc(directory, filename, exclude);
ajListSort(listTestFiles, ajStrVcmp);
nfiles = ajListToarray(listTestFiles, &testFiles);
if(!nfiles)
ajDie("No input files in '%S' matched filename '%S'",
directory, filename);
embDbiLogHeader(logfile, dbname, release, datestr,
indexdir, maxindex);
embDbiLogFields(logfile, fields, nfields);
embDbiLogSource(logfile, directory, filename, exclude,
(AjPStr*) testFiles, nfiles);
embDbiLogCmdline(logfile);
AJCNEW0(divfiles, nfiles);
/*
** process each input file, one at a time
*/
jfile = 0;
for(ifile=0; ifile < nfiles; ifile++)
{
curfilename = (AjPStr) testFiles[ifile];
if(!dbiblast_blastopenlib(curfilename,
usesrc, blastv, dbtype, &db))
continue; /* could be the wrong file type with "*.*" */
ajDebug("processing filename '%S' ...\n", curfilename);
ajDebug("processing file '%S' ...\n", db->TFile->Name);
ajStrAssignS(&divfiles[jfile], db->TFile->Name);
ajFilenameTrimPath(&divfiles[jfile]);
if(ajStrGetLen(divfiles[jfile]) >= maxfilelen)
maxfilelen = ajStrGetLen(divfiles[jfile]) + 1;
if(systemsort) /* elistfile for entries, alist for fields */
elistfile = embDbiSortOpen(alistfile, jfile,
dbname, fields, nfields);
idCountFile = 0;
for(i=0;i<nfields;i++)
countField[i] = 0;
while((entry=dbiblast_nextblastentry(db, jfile,
idformat, systemsort,
fields,
maxFieldLen,
&maxidlen, countField,
elistfile, alistfile)))
{
idCountFile++;
if(!systemsort) /* save the entry data in lists */
{
embDbiMemEntry(idlist, fieldList, nfields, entry, jfile);
}
}
idCount += idCountFile;
if(systemsort)
{
embDbiSortClose(&elistfile, alistfile, nfields);
/* lost the entry, so can't free it :-) */
}
embDbiLogFile(logfile, curfilename, idCountFile, fields,
countField, nfields);
dbiblast_dbfree(&db);
jfile++;
}
nfiles = jfile;
/*
** write the division.lkp file
*/
embDbiWriteDivision(indexdir, dbname, release, date,
maxfilelen, nfiles, divfiles, NULL);
/*
** Write the entryname.idx index
*/
ajStrAssignC(&tmpfname, "entrynam.idx");
entFile = ajFileNewOutNamePathS(tmpfname, indexdir);
recsize = maxidlen+10;
filesize = 300 + (idCount*(ajint)recsize);
embDbiHeader(entFile, filesize, idCount, recsize, dbname, release, date);
if(systemsort)
idDone = embDbiSortWriteEntry(entFile, maxidlen,
dbname, nfiles, cleanup, sortopt);
else /* save entries in entryIds array */
{
idDone = embDbiMemWriteEntry(entFile, maxidlen,
idlist, &entryIds);
if(idDone != idCount)
ajFatal("Duplicates not allowed for in-memory processing");
}
embDbiHeaderSize(entFile, 300+(idDone*(ajint)recsize), idDone);
ajFileClose(&entFile);
/*
** Write the fields index files
*/
for(ifield=0; ifield < nfields; ifield++)
{
if(maxindex)
maxlen = maxindex;
else
{
if(maxFieldLen[ifield] >= 0)
maxlen = maxFieldLen[ifield];
else
maxlen = - maxFieldLen[ifield];
}
if(systemsort)
fieldTot[ifield] = embDbiSortWriteFields(dbname, release,
date, indexdir,
fields[ifield], maxlen,
nfiles, idCount,
cleanup, sortopt);
else
fieldTot[ifield] = embDbiMemWriteFields(dbname, release,
date, indexdir,
fields[ifield], maxlen,
fieldList[ifield],
entryIds);
}
embDbiLogFinal(logfile,maxindex, maxFieldLen, fields, fieldTot,
nfields, nfiles, idDone, idCount);
if(systemsort)
embDbiRmEntryFile(dbname, cleanup);
ajListMap(idlist, embDbiEntryDelMap, NULL);
ajListFree(&idlist);
AJFREE(entryIds);
ajStrDelarray(&fields);
for(i=0;i<nfields;i++)
{
if(systemsort)
{
ajFileClose(&alistfile[i]);
}
else
{
ajListMap(fieldList[i], embDbiFieldDelMap, NULL);
ajListFree(&fieldList[i]);
}
}
AJFREE(alistfile);
AJFREE(fieldList);
ajStrDel(&version);
ajStrDel(&seqtype);
ajFileClose(&elistfile);
for(i=0;i<nfiles;i++)
{
ajStrDel(&divfiles[i]);
}
AJFREE(countField);
AJFREE(fieldTot);
ajStrDel(&dbname);
ajStrDel(&release);
ajStrDel(&datestr);
ajStrDel(&sortopt);
ajStrDel(&directory);
ajStrDel(&indexdir);
ajStrDel(&filename);
ajStrDel(&exclude);
ajStrDel(&idformat);
ajStrDel(&tmpfname);
AJFREE(maxFieldLen);
ajFileClose(&logfile);
ajListstrFreeData(&listTestFiles);
ajStrDel(&t);
ajStrDel(&id);
ajStrDel(&acc);
ajStrDel(&hline);
ajStrDel(&tmpdes);
ajStrDel(&tmpfd);
ajStrDel(&tmpgi);
ajStrDel(&tmpdb);
ajStrDel(&tmpac);
ajStrDel(&tmpsv);
ajRegFree(&wrdexp);
embDbiEntryDel(&dbiblastEntry);
if(fdl)
{
for(i=0; i < nfields; i++)
ajListFree(&fdl[i]);
AJFREE(fdl);
}
for(i=0;i<nfiles;i++)
{
ajStrDel(&divfiles[i]);
}
AJFREE(divfiles);
AJFREE(testFiles);
embExit();
return 0;
}
AjPMatrix ajMatrixNewFile(const AjPStr filename)
{
AjPMatrix ret = NULL;
AjPStr buffer = NULL;
const AjPStr tok = NULL;
AjPStr firststring = NULL;
AjPStr *orderstring = NULL;
AjPFile file = NULL;
AjBool first = ajTrue;
const char *ptr = NULL;
ajint **matrix = NULL;
ajint minval = -1;
ajint i = 0;
ajint l = 0;
ajint k = 0;
ajint cols = 0;
ajint rows = 0;
ajint *templine = NULL;
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();
file = ajDatafileNewInNameS(filename);
if(!file)
{
ajStrDel(&firststring);
ajListFree(&rlabel_list);
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 = ajMatrixNewAsym(orderstring, cols,
rlabel_arr, rows,
filename);
matrix = ret->Matrix;
}
else
{
ajFmtScanC(ptr, "%S", &firststring);
/* JISON 19/7/4
k = ajSeqcvtGetCodeK(ret->Cvt, ajStrGetCharFirst(firststring)); */
k = ajSeqcvtGetCodeS(ret->Cvt, firststring);
/*
** cols+1 is used below because 2nd and subsequent lines have
** one more string in them (the residue label)
*/
templine = ajArrIntLine(buffer,delimstr,2,cols+1);
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);
for(i=0; i<rows; i++)
ajStrDel(&rlabel_arr[i]);
AJFREE(rlabel_arr);
ajListFree(&rlabel_list);
return ret;
}
void emboss_copy(AjPSeqset seqset, char ***retseqs, AINFO *info)
{
ajint n;
ajint maxlen;
ajint len;
char **seqs;
const AjPSeq seq = NULL;
ajint i=0;
const AjPStr fmt=NULL;
const char *p=NULL;
char c='\0';
/*
char *q=NULL;
AjPSelexseq sqdata=NULL;
AjPSelexdata sdata=NULL;
*/
ajint cnt=0;
info->name = NULL;
info->rf=NULL;
info->cs=NULL;
info->desc=NULL;
info->acc=NULL;
info->au=NULL;
info->flags=0;
AjPStr tmpstr = NULL;
ajSeqsetFill(seqset);
fmt = ajSeqsetGetFormat(seqset);
n = ajSeqsetGetSize(seqset);
ajSeqsetFmtUpper(seqset);
maxlen = ajSeqsetGetLen(seqset);
/* First allocate and copy sequences */
AJCNEW0(seqs,n);
for(i=0; i<n; ++i)
{
seqs[i] = ajCharNewRes(maxlen+1);
strcpy(seqs[i],ajSeqGetSeqC(ajSeqsetGetseqSeq(seqset,i)));
}
info->sqinfo = (SQINFO *) calloc (sizeof(SQINFO), n);
for(i=0; i<n; ++i)
{
info->sqinfo[i].flags = 0;
strcpy(info->sqinfo[i].name,"");
strcpy(info->sqinfo[i].id,"");
strcpy(info->sqinfo[i].acc,"");
strcpy(info->sqinfo[i].desc,"");
info->sqinfo[i].len = 0;
info->sqinfo[i].start = 0;
info->sqinfo[i].stop = 0;
info->sqinfo[i].olen = 0;
info->sqinfo[i].type = 0;
info->sqinfo[i].ss = NULL;
info->sqinfo[i].sa =NULL;
}
AJCNEW0(info->wgt,n);
for(i=0; i<n; ++i)
{
info->sqinfo[i].flags = 0;
info->wgt[i] = ajSeqsetGetseqWeight(seqset,i);
}
info->nseq = n;
info->alen = maxlen;
for(i=0; i<n; ++i)
{
seq = ajSeqsetGetseqSeq(seqset,i);
if((len=ajStrGetLen(ajSeqGetNameS(seq))))
{
if(len>= SQINFO_NAMELEN)
len = SQINFO_NAMELEN - 1;
ajStrAssignSubS(&tmpstr, ajSeqGetNameS(seq), 0, len);
strcpy(info->sqinfo[i].id,ajStrGetPtr(tmpstr));
info->sqinfo[i].flags |= SQINFO_ID;
strcpy(info->sqinfo[i].name,ajStrGetPtr(tmpstr));
info->sqinfo[i].flags |= SQINFO_NAME;
}
if((len=ajStrGetLen(ajSeqGetAccS(seq))))
{
if(len>= SQINFO_NAMELEN)
len = SQINFO_NAMELEN - 1;
ajStrAssignSubS(&tmpstr, ajSeqGetAccS(seq), 0, len);
strcpy(info->sqinfo[i].acc,ajStrGetPtr(tmpstr));
info->sqinfo[i].flags |= SQINFO_ACC;
}
}
seq = ajSeqsetGetseqSeq(seqset,0);
info->cs = ajCharNewS(ajSeqGetSeqS(seq));
info->name = ajCharNewS(ajSeqGetNameS(seq));
info->acc = ajCharNewS(ajSeqGetAccS(seq));
info->desc = ajCharNewS(ajSeqGetDescS(seq));
info->rf = ajCharNewS(ajSeqGetSeqS(seq));
/*
info->rf = ajCharNewS(seq);
len = ajStrGetLen(seq->Selexdata->name);
info->name = ajCharNewRes(len+1);
strcpy(info->name,ajStrGetPtr(seq->Selexdata->name));
len = ajStrGetLen(seq->Selexdata->de);
info->desc = ajCharNewRes(len+1);
sdata = seq->Selexdata;
strcpy(info->desc,ajStrGetPtr(sdata->de));
len = ajStrGetLen(sdata->ac);
info->acc = ajCharNewRes(len+1);
strcpy(info->acc,ajStrGetPtr(sdata->ac));
len = ajStrGetLen(sdata->au);
info->au = ajCharNewRes(len+1);
strcpy(info->au,ajStrGetPtr(sdata->au));
if(sdata->tc[0] || sdata->tc[1])
{
info->flags |= AINFO_TC;
info->tc1 = sdata->tc[0];
info->tc2 = sdata->tc[1];
}
if(sdata->nc[0] || sdata->nc[1])
{
info->flags |= AINFO_NC;
info->nc1 = sdata->nc[0];
info->nc2 = sdata->nc[1];
}
if(sdata->ga[0] || sdata->ga[1])
{
info->flags |= AINFO_GA;
info->ga1 = sdata->ga[0];
info->ga2 = sdata->ga[1];
}
for(i=0;i<n;++i)
{
seq = ajSeqsetGetseqSeq(seqset,i);
sqdata = seq->Selexdata->sq;
if((len=ajStrGetLen(sqdata->name)))
{
if(len<64)
strcpy(info->sqinfo[i].name,ajStrGetPtr(sqdata->name));
else
strncpy(info->sqinfo[i].name,ajStrGetPtr(sqdata->name),63);
info->sqinfo[i].name[63]='\0';
info->sqinfo[i].flags |= SQINFO_NAME;
}
/ *
if((len=ajStrGetLen(sqdata->id)))
{
if(len<64)
strcpy(info->sqinfo[i].id,ajStrGetPtr(sqdata->id));
else
strncpy(info->sqinfo[i]->id,ajStrGetPtr(sqdata->id),63);
info->sqinfo[i].id[63]='\0';
info->sqinfo[i].flags |= SQINFO_ID;
}
* /
strcpy(info->sqinfo[i].id,info->sqinfo[i].name);
info->sqinfo[i].flags |= SQINFO_ID;
if((len=ajStrGetLen(sqdata->ac)))
{
if(len<64)
strcpy(info->sqinfo[i].acc,ajStrGetPtr(sqdata->ac));
else
strncpy(info->sqinfo[i].acc,ajStrGetPtr(sqdata->ac),63);
info->sqinfo[i].acc[63]='\0';
info->sqinfo[i].flags |= SQINFO_ACC;
}
if((len=ajStrGetLen(sqdata->de)))
{
if(len<127)
strcpy(info->sqinfo[i].desc,ajStrGetPtr(sqdata->de));
else
strncpy(info->sqinfo[i].desc,ajStrGetPtr(sqdata->de),127);
info->sqinfo[i].desc[127]='\0';
info->sqinfo[i].flags |= SQINFO_DESC;
}
if(sqdata->start || sqdata->stop || sqdata ->len)
{
info->sqinfo[i].start = sqdata->start;
info->sqinfo[i].stop = sqdata->stop;
info->sqinfo[i].olen = sqdata->len;
info->sqinfo[i].flags |= SQINFO_START;
info->sqinfo[i].flags |= SQINFO_STOP;
info->sqinfo[i].flags |= SQINFO_OLEN;
}
if(ajStrGetLen(seq->Selexdata->ss))
{
info->sqinfo[i].ss = ajCharNewRes(maxlen+1);
p = ajStrGetPtr(seq->Selexdata->ss);
q = info->sqinfo[i].ss;
while((c==*p))
{
if(c=='.' || c==' ' || c=='_' || c=='-')
*q++ = c;
++p;
}
*q = '\0';
info->sqinfo[i].flags |= SQINFO_SS;
}
}
}
/ *
}
*/
for(i=0; i<n; ++i)
{
info->sqinfo[i].type = kOtherSeq;
if(ajSeqsetIsDna(seqset))
info->sqinfo[i].type = kDNA;
if(ajSeqsetIsRna(seqset))
info->sqinfo[i].type = kRNA;
if(ajSeqsetIsProt(seqset))
info->sqinfo[i].type = kAmino;
info->sqinfo[i].flags |= SQINFO_TYPE;
seq = ajSeqsetGetseqSeq(seqset,i);
p = ajSeqGetSeqC(seq);
cnt = 0;
while((c=*p))
{
if(!(c=='.' || c==' ' || c=='_' || c=='-' || c=='~'))
++cnt;
++p;
}
info->sqinfo[i].len = cnt;
info->sqinfo[i].flags |= SQINFO_LEN;
}
*retseqs = seqs;
ajStrDel(&tmpstr);
return;
}
EmbPPropMolwt* embPropEmolwtRead(AjPFile mfptr)
{
AjPStr line = NULL;
AjPStr token = NULL;
AjBool firstline;
const char *p;
ajuint i;
ajint n;
EmbPPropMolwt *ret;
line = ajStrNew();
token = ajStrNew();
firstline = ajTrue;
AJCNEW0(ret,EMBPROPSIZE+2);
for(i=0; i < EMBPROPSIZE+2; ++i)
AJNEW0(ret[i]);
while(ajReadline(mfptr, &line))
{
ajStrRemoveWhiteExcess(&line);
p = ajStrGetPtr(line);
if(*p=='#' || *p=='!' || !*p)
continue;
if(firstline)
{
if(!ajStrPrefixC(line,"Mol"))
ajFatal("Incorrect format molwt file: '%S'", line);
firstline = ajFalse;
continue;
}
ajFmtScanS(line,"%S",&token);
ajStrFmtUpper(&token);
if(ajStrGetLen(token) != 1)
{
if(ajStrPrefixC(token,"HYDROGEN"))
{
if(ajFmtScanS(line,"%*s%lf%lf",
&ret[EMBPROPHINDEX]->average,
&ret[EMBPROPHINDEX]->mono) != 2)
ajFatal("Bad format hydrogen data line");
}
else if(ajStrPrefixC(token,"OXYGEN"))
{
if(ajFmtScanS(line,"%*s%lf%lf",
&ret[EMBPROPOINDEX]->average,
&ret[EMBPROPOINDEX]->mono) != 2)
ajFatal("Bad format oxygen data line");
}
else if(ajStrPrefixC(token,"WATER"))
{
if(ajFmtScanS(line,"%*s%lf%lf",
&ret[EMBPROPWINDEX]->average,
&ret[EMBPROPWINDEX]->mono) != 2)
ajFatal("Bad format water data line");
}
else
ajFatal("Unknown molwt token %S",token);
continue;
}
i = ajBasecodeToInt((ajint) *ajStrGetPtr(token));
if(i == 27)
ajFatal("Molwt file line doesn't begin with a single A->Z (%S)",
line);
n = ajFmtScanS(line,"%*s%lf%lf",
&ret[i]->average,
&ret[i]->mono);
if(n != 2)
ajFatal("Only %d columns in amino file - expected %d",n,3);
}
ajStrDel(&line);
ajStrDel(&token);
return ret;
}
int main(int argc, char **argv)
{
AjPSeqall queryseqs;
AjPSeqset targetseqs;
AjPSeq queryseq;
const AjPSeq targetseq;
AjPStr queryaln = 0;
AjPStr targetaln = 0;
AjPFile errorf;
AjBool show = ajFalse;
const char *queryseqc;
const char *targetseqc;
AjPMatrixf matrix;
AjPSeqCvt cvt = 0;
float **sub;
ajint *compass = NULL;
float *path = NULL;
float gapopen;
float gapextend;
float score;
float minscore;
ajuint j, k;
ajint querystart = 0;
ajint targetstart = 0;
ajint queryend = 0;
ajint targetend = 0;
ajint width = 0;
AjPTable kmers = 0;
ajint wordlen = 6;
ajint oldmax = 0;
ajint newmax = 0;
ajuint ntargetseqs;
ajuint nkmers;
AjPAlign align = NULL;
EmbPWordMatch maxmatch; /* match with maximum score */
/* Cursors for the current sequence being scanned,
** i.e., until which location it was scanned.
** Separate cursor/location entries for each sequence in the seqset.
*/
ajuint* lastlocation;
EmbPWordRK* wordsw = NULL;
AjPList* matchlist = NULL;
embInit("supermatcher", argc, argv);
matrix = ajAcdGetMatrixf("datafile");
queryseqs = ajAcdGetSeqall("asequence");
targetseqs= ajAcdGetSeqset("bsequence");
gapopen = ajAcdGetFloat("gapopen");
gapextend = ajAcdGetFloat("gapextend");
wordlen = ajAcdGetInt("wordlen");
align = ajAcdGetAlign("outfile");
errorf = ajAcdGetOutfile("errorfile");
width = ajAcdGetInt("width"); /* width for banded Smith-Waterman */
minscore = ajAcdGetFloat("minscore");
gapopen = ajRoundFloat(gapopen, 8);
gapextend = ajRoundFloat(gapextend, 8);
sub = ajMatrixfGetMatrix(matrix);
cvt = ajMatrixfGetCvt(matrix);
embWordLength(wordlen);
/* seqset sequence is the reference sequence for SAM format */
ajAlignSetRefSeqIndx(align, 1);
ajSeqsetTrim(targetseqs);
ntargetseqs = ajSeqsetGetSize(targetseqs);
AJCNEW0(matchlist, ntargetseqs);
/* get tables of words */
for(k=0;k<ntargetseqs;k++)
{
targetseq = ajSeqsetGetseqSeq(targetseqs, k);
embWordGetTable(&kmers, targetseq);
ajDebug("Number of distinct kmers found so far: %d\n",
ajTableGetLength(kmers));
}
AJCNEW0(lastlocation, ntargetseqs);
if(ajTableGetLength(kmers)<1)
ajErr("no kmers found");
nkmers = embWordRabinKarpInit(kmers, &wordsw, wordlen, targetseqs);
while(ajSeqallNext(queryseqs,&queryseq))
{
ajSeqTrim(queryseq);
queryaln = ajStrNewRes(1+ajSeqGetLen(queryseq));
ajDebug("Read '%S'\n", ajSeqGetNameS(queryseq));
for(k=0;k<ntargetseqs;k++)
{
lastlocation[k]=0;
matchlist[k] = ajListstrNew();
}
embWordRabinKarpSearch(ajSeqGetSeqS(queryseq), targetseqs,
(const EmbPWordRK*)wordsw, wordlen, nkmers,
matchlist, lastlocation, ajFalse);
for(k=0;k<ajSeqsetGetSize(targetseqs);k++)
{
targetseq = ajSeqsetGetseqSeq(targetseqs, k);
ajDebug("Processing '%S'\n", ajSeqGetNameS(targetseq));
if(ajListGetLength(matchlist[k])==0)
{
ajFmtPrintF(errorf,
"No wordmatch start points for "
"%s vs %s. No alignment\n",
ajSeqGetNameC(queryseq),ajSeqGetNameC(targetseq));
embWordMatchListDelete(&matchlist[k]);
continue;
}
/* only the maximum match is used as seed
* (if there is more than one location with the maximum match
* only the first one is used)
* TODO: we should add a new option to make above limit optional
*/
maxmatch = embWordMatchFirstMax(matchlist[k]);
supermatcher_findendpoints(maxmatch,targetseq, queryseq,
&targetstart, &querystart,
&targetend, &queryend);
targetaln=ajStrNewRes(1+ajSeqGetLen(targetseq));
queryseqc = ajSeqGetSeqC(queryseq);
targetseqc = ajSeqGetSeqC(targetseq);
ajStrAssignC(&queryaln,"");
ajStrAssignC(&targetaln,"");
ajDebug("++ %S v %S start:%d %d end:%d %d\n",
ajSeqGetNameS(targetseq), ajSeqGetNameS(queryseq),
targetstart, querystart, targetend, queryend);
newmax = (targetend-targetstart+2)*width;
if(newmax > oldmax)
{
AJCRESIZE0(path,oldmax,newmax);
AJCRESIZE0(compass,oldmax,newmax);
oldmax=newmax;
ajDebug("++ memory re/allocation for path/compass arrays"
" to size: %d\n", newmax);
}
else
{
AJCSET0(path,newmax);
AJCSET0(compass,newmax);
}
ajDebug("Calling embAlignPathCalcSWFast "
"%d..%d [%d/%d] %d..%d [%d/%d] width:%d\n",
querystart, queryend, (queryend - querystart + 1),
ajSeqGetLen(queryseq),
targetstart, targetend, (targetend - targetstart + 1),
ajSeqGetLen(targetseq),
width);
score = embAlignPathCalcSWFast(&targetseqc[targetstart],
&queryseqc[querystart],
targetend-targetstart+1,
queryend-querystart+1,
0,width,
gapopen,gapextend,
path,sub,cvt,
compass,show);
if(score>minscore)
{
embAlignWalkSWMatrixFast(path,compass,gapopen,gapextend,
targetseq,queryseq,
&targetaln,&queryaln,
targetend-targetstart+1,
queryend-querystart+1,
0,width,
&targetstart,&querystart);
if(!ajAlignFormatShowsSequences(align))
{
ajAlignDefineCC(align, ajStrGetPtr(targetaln),
ajStrGetPtr(queryaln),
ajSeqGetNameC(targetseq),
ajSeqGetNameC(queryseq));
ajAlignSetScoreR(align, score);
}
else
{
ajDebug(" queryaln:%S \ntargetaln:%S\n",
queryaln,targetaln);
embAlignReportLocal(align,
queryseq, targetseq,
queryaln, targetaln,
querystart, targetstart,
gapopen, gapextend,
score, matrix,
1 + ajSeqGetOffset(queryseq),
1 + ajSeqGetOffset(targetseq)
);
}
ajAlignWrite(align);
ajAlignReset(align);
}
ajStrDel(&targetaln);
embWordMatchListDelete(&matchlist[k]);
}
ajStrDel(&queryaln);
}
for(k=0;k<nkmers;k++)
{
AJFREE(wordsw[k]->seqindxs);
AJFREE(wordsw[k]->nSeqMatches);
for(j=0;j<wordsw[k]->nseqs;j++)
AJFREE(wordsw[k]->locs[j]);
AJFREE(wordsw[k]->nnseqlocs);
AJFREE(wordsw[k]->locs);
AJFREE(wordsw[k]);
}
embWordFreeTable(&kmers);
if(!ajAlignFormatShowsSequences(align))
ajMatrixfDel(&matrix);
AJFREE(path);
AJFREE(compass);
AJFREE(kmers);
AJFREE(wordsw);
AJFREE(matchlist);
AJFREE(lastlocation);
ajAlignClose(align);
ajAlignDel(&align);
ajSeqallDel(&queryseqs);
ajSeqDel(&queryseq);
ajSeqsetDel(&targetseqs);
ajFileClose(&errorf);
embExit();
return 0;
}
int main(int argc, char **argv)
{
AjPSeq seq;
AjPGraph graph = 0;
AjPFile outf = NULL;
AjPFile file = NULL;
AjPStr buffer = NULL;
float twist[4][4][4];
float roll[4][4][4];
float tilt[4][4][4];
float rbend;
float rcurve;
float bendscale;
float curvescale;
float twistsum = (float) 0.0;
float pi = (float) 3.14159;
float pifac = (pi/(float) 180.0);
float pi2 = pi/(float) 2.0;
ajint *iseq = NULL;
float *x;
float *y;
float *xave;
float *yave;
float *curve;
float *bend;
const char *ptr;
ajint i;
ajint ii;
ajint k;
ajint j;
char residue[2];
float maxbend;
float minbend;
float bendfactor;
float maxcurve;
float mincurve;
float curvefactor;
float fxp;
float fyp;
float yincr;
float yy1;
ajint ixlen;
ajint iylen;
ajint ixoff;
ajint iyoff;
float ystart;
float defheight;
float currentheight;
ajint count;
ajint portrait = 0;
ajint title = 0;
ajint numres;
ajint ibeg;
ajint iend;
ajint ilen;
AjPStr sstr = NULL;
ajint ipos;
float dx;
float dy;
float rxsum;
float rysum;
float yp1;
float yp2;
double td;
embInit("banana", argc, argv);
seq = ajAcdGetSeq("sequence");
file = ajAcdGetDatafile("anglesfile");
outf = ajAcdGetOutfile("outfile");
graph = ajAcdGetGraph("graph");
numres = ajAcdGetInt("residuesperline");
ibeg = ajSeqGetBegin(seq);
iend = ajSeqGetEnd(seq);
ajStrAssignSubS(&sstr, ajSeqGetSeqS(seq), ibeg-1, iend-1);
ilen = ajStrGetLen(sstr);
AJCNEW0(iseq,ilen+1);
AJCNEW0(x,ilen+1);
AJCNEW0(y,ilen+1);
AJCNEW0(xave,ilen+1);
AJCNEW0(yave,ilen+1);
AJCNEW0(curve,ilen+1);
AJCNEW0(bend,ilen+1);
ptr= ajStrGetPtr(sstr);
for(ii=0;ii<ilen;ii++)
{
if(*ptr=='A' || *ptr=='a')
iseq[ii+1] = 0;
else if(*ptr=='T' || *ptr=='t')
iseq[ii+1] = 1;
else if(*ptr=='G' || *ptr=='g')
iseq[ii+1] = 2;
else if(*ptr=='C' || *ptr=='c')
iseq[ii+1] = 3;
else
ajErr("%c is not an ATCG hence not valid",*ptr);
ptr++;
}
if(!file)
ajErr("Banana failed to open angle file");
ajReadline(file,&buffer); /* 3 junk lines */
ajReadline(file,&buffer);
ajReadline(file,&buffer);
for(k=0;k<4;k++)
for(ii=0;ii<4;ii++)
{
if(ajReadline(file,&buffer))
{
sscanf(ajStrGetPtr(buffer),"%f,%f,%f,%f",
&twist[ii][0][k],&twist[ii][1][k],&twist[ii][2][k],
&twist[ii][3][k]);
}
else
ajErr("Error reading angle file");
for(j=0;j<4;j++)
twist[ii][j][k] *= pifac;
}
for(k=0;k<4;k++)
for(ii=0;ii<4;ii++)
if(ajReadline(file,&buffer))
{
sscanf(ajStrGetPtr(buffer),"%f,%f,%f,%f",&roll[ii][0][k],
&roll[ii][1][k],&roll[ii][2][k],&roll[ii][3][k]);
}
else
ajErr("Error reading angle file");
for(k=0;k<4;k++)
for(ii=0;ii<4;ii++)
if(ajReadline(file,&buffer))
sscanf(ajStrGetPtr(buffer),"%f,%f,%f,%f",&tilt[ii][0][k],
&tilt[ii][1][k],&tilt[ii][2][k],&tilt[ii][3][k]);
else
ajErr("Error reading angle file");
if(ajReadline(file,&buffer))
sscanf(ajStrGetPtr(buffer),"%f,%f,%f,%f",&rbend,&rcurve,
&bendscale,&curvescale);
else
ajErr("Error reading angle file");
ajFileClose(&file);
ajStrDel(&buffer);
for(ii=1;ii<ilen-1;ii++)
{
twistsum += twist[iseq[ii]][iseq[ii+1]][iseq[ii+2]];
dx = (roll[iseq[ii]][iseq[ii+1]][iseq[ii+2]]*sinfban(twistsum)) +
(tilt[iseq[ii]][iseq[ii+1]][iseq[ii+2]]*sinfban(twistsum-pi2));
dy = roll[iseq[ii]][iseq[ii+1]][iseq[ii+2]]*cosfban(twistsum) +
tilt[iseq[ii]][iseq[ii+1]][iseq[ii+2]]*cosfban(twistsum-pi2);
x[ii+1] = x[ii]+dx;
y[ii+1] = y[ii]+dy;
}
for(ii=6;ii<ilen-6;ii++)
{
rxsum = 0.0;
rysum = 0.0;
for(k=-4;k<=4;k++)
{
rxsum+=x[ii+k];
rysum+=y[ii+k];
}
rxsum+=(x[ii+5]*(float)0.5);
rysum+=(y[ii+5]*(float)0.5);
rxsum+=(x[ii-5]*(float)0.5);
rysum+=(y[ii-5]*(float)0.5);
xave[ii] = rxsum*(float)0.1;
yave[ii] = rysum*(float)0.1;
}
for(i=(ajint)rbend+1;i<=ilen-(ajint)rbend-1;i++)
{
td = sqrt(((x[i+(ajint)rbend]-x[i-(ajint)rbend])*
(x[i+(ajint)rbend]-x[i-(ajint)rbend])) +
((y[i+(ajint)rbend]-y[i-(ajint)rbend])*
(y[i+(ajint)rbend]-y[i-(ajint)rbend])));
bend[i] = (float) td;
bend[i]*=bendscale;
}
for(i=(ajint)rcurve+6;i<=ilen-(ajint)rcurve-6;i++)
{
td = sqrt(((xave[i+(ajint)rcurve]-
xave[i-(ajint)rcurve])*(xave[i+(ajint)rcurve]-
xave[i-(ajint)rcurve]))+
((yave[i+(ajint)rcurve]-yave[i-(ajint)rcurve])*
(yave[i+(ajint)rcurve]-yave[i-(ajint)rcurve])));
curve[i] = (float) td;
}
if(outf)
{
ajFmtPrintF(outf,"Base Bend Curve\n");
ptr = ajStrGetPtr(sstr);
for(ii=1;ii<=ilen;ii++)
{
ajFmtPrintF(outf,"%c %6.1f %6.1f\n",
*ptr, bend[ii], curve[ii]);
ptr++;
}
ajFileClose(&outf);
}
if(graph)
{
maxbend = minbend = 0.0;
maxcurve = mincurve = 0.0;
for(ii=1;ii<=ilen;ii++)
{
if(bend[ii] > maxbend)
maxbend = bend[ii];
if(bend[ii] < minbend)
minbend = bend[ii];
if(curve[ii] > maxcurve)
maxcurve = curve[ii];
if(curve[ii] < mincurve)
mincurve = curve[ii];
}
ystart = 75.0;
ajGraphAppendTitleS(graph, ajSeqGetUsaS(seq));
ajGraphicsSetPagesize(960, 768);
ajGraphOpenWin(graph,(float)-1.0, (float)numres+(float)10.0,
(float)0.0, ystart+(float)5.0);
ajGraphicsGetParamsPage(&fxp,&fyp,&ixlen,&iylen,&ixoff,&iyoff);
if(portrait)
{
ixlen = 768;
iylen = 960;
}
else
{
ixlen = 960;
iylen = 768;
}
ajGraphicsGetCharsize(&defheight,¤theight);
if(!currentheight)
{
defheight = currentheight = (float) 4.440072;
currentheight = defheight *
((float)ixlen/ ((float)(numres)*(currentheight+(float)1.0)))
/currentheight;
}
ajGraphicsSetCharscale(((float)ixlen/((float)(numres)*
(currentheight+(float)1.0)))/
currentheight);
ajGraphicsGetCharsize(&defheight,¤theight);
yincr = (currentheight + (float)3.0)*(float)0.3;
if(!title)
yy1 = ystart;
else
yy1 = ystart-(float)5.0;
count = 1;
residue[1]='\0';
bendfactor = (3*yincr)/maxbend;
curvefactor = (3*yincr)/maxcurve;
ptr = ajStrGetPtr(sstr);
yy1 = yy1-(yincr*((float)5.0));
for(ii=1;ii<=ilen;ii++)
{
if(count > numres)
{
yy1 = yy1-(yincr*((float)5.0));
if(yy1<1.0)
{
if(!title)
yy1=ystart;
else
yy1 = ystart-(float)5.0;
yy1 = yy1-(yincr*((float)5.0));
ajGraphNewpage(graph,AJFALSE);
}
count = 1;
}
residue[0] = *ptr;
ajGraphicsDrawposTextAtend((float)(count)+(float)2.0,yy1,residue);
if(ii>1 && ii < ilen)
{
yp1 = yy1+yincr + (bend[ii]*bendfactor);
yp2 = yy1+yincr + (bend[ii+1]*bendfactor);
ajGraphicsDrawposLine((float)count+(float)1.5,yp1,
(float)(count)+(float)2.5,yp2);
}
ipos = ilen-(ajint)rcurve-7;
if(ipos < 0)
ipos = 0;
if(ii>rcurve+5 && ii<ipos)
{
yp1 = yy1+yincr + (curve[ii]*curvefactor);
yp2 = yy1+yincr + (curve[ii+1]*curvefactor);
ajGraphicsDrawposLine((float)count+(float)1.7,yp1,
(float)(count)+(float)2.3,yp2);
}
ajGraphicsDrawposLine((float)count+(float)1.5,yy1+yincr,
(float)(count)+(float)2.5,yy1+yincr);
count++;
ptr++;
}
ajGraphicsClose();
}
AJFREE(iseq);
AJFREE(x);
AJFREE(y);
AJFREE(xave);
AJFREE(yave);
AJFREE(curve);
AJFREE(bend);
ajStrDel(&sstr);
ajSeqDel(&seq);
ajFileClose(&file);
ajFileClose(&outf);
ajGraphxyDel(&graph);
embExit();
return 0;
}
int main(int argc, char **argv)
{
/* Global details */
AjBool explain_flag;
AjBool file_flag;
AjPStr* task;
AjBool do_primer;
AjBool do_hybrid;
ajint num_return;
ajint first_base_index;
/* "Sequence" Input Tags */
AjPSeqall sequence;
AjPRange included_region;
AjPRange target;
AjPRange excluded_region;
AjPStr left_input;
AjPStr right_input;
/* Primer details */
AjBool pick_anyway;
AjPFile mispriming_library;
float max_mispriming;
float pair_max_mispriming;
ajint gc_clamp;
ajint opt_size;
ajint min_size;
ajint max_size;
float opt_tm;
float min_tm;
float max_tm;
float max_diff_tm;
float opt_gc_percent;
float min_gc;
float max_gc;
float salt_conc;
float dna_conc;
ajint num_ns_accepted;
float self_any;
float self_end;
ajint max_poly_x;
/* Sequence Quality. These are not (yet) implemented */
/*
AjPFile sequence_quality;
ajint min_quality;
ajint min_end_quality;
ajint quality_range_min;
ajint quality_range_max;
*/
/* Product details */
ajint product_opt_size;
AjPRange product_size_range;
float product_opt_tm;
float product_min_tm;
float product_max_tm;
/* Objective Function Penalty Weights for Primers */
float max_end_stability;
/* these are not (yet) implemented */
/*
float inside_penalty;
float outside_penalty;
*/
/* Primer penalties */
/* these are not (yet) implemented */
/* Internal Oligo "Sequence" Input Tags */
AjPRange internal_oligo_excluded_region;
/* Internal Oligo "Global" Input Tags */
AjPStr internal_oligo_input;
ajint internal_oligo_opt_size;
ajint internal_oligo_min_size;
ajint internal_oligo_max_size;
float internal_oligo_opt_tm;
float internal_oligo_min_tm;
float internal_oligo_max_tm;
float internal_oligo_opt_gc_percent;
float internal_oligo_min_gc;
float internal_oligo_max_gc;
float internal_oligo_salt_conc;
float internal_oligo_dna_conc;
float internal_oligo_self_any;
float internal_oligo_self_end;
ajint internal_oligo_max_poly_x;
AjPFile internal_oligo_mishyb_library;
float internal_oligo_max_mishyb;
/*
ajint internal_oligo_min_quality;
*/
/* Internal Oligo penalties */
/* these are not (yet) implemented */
/* EMBOSS-wrapper-specific stuff */
AjPFile outfile;
/* other variables */
AjPStr result = NULL;
AjPStr strand = NULL;
AjPStr substr = NULL;
AjPSeq seq = NULL;
ajint begin = 0;
ajint end;
FILE* stream;
AjPStr taskstr = NULL;
const AjPStr program = NULL;
/* pipe variables */
int *pipeto; /* pipe to feed the exec'ed program input */
int *pipefrom; /* pipe to get the exec'ed program output */
embInit("eprimer3", argc, argv);
/* Global details */
explain_flag = ajAcdGetBoolean("explainflag");
file_flag = ajAcdGetBoolean("fileflag");
task = ajAcdGetList("task");
do_primer = ajAcdGetToggle("primer");
do_hybrid = ajAcdGetToggle("hybridprobe");
num_return = ajAcdGetInt("numreturn");
first_base_index = ajAcdGetInt("firstbaseindex");
/* "Sequence" Input Tags */
sequence = ajAcdGetSeqall("sequence");
included_region = ajAcdGetRange("includedregion");
target = ajAcdGetRange("targetregion");
excluded_region = ajAcdGetRange("excludedregion");
left_input = ajAcdGetString("forwardinput");
right_input = ajAcdGetString("reverseinput");
/* Primer details */
pick_anyway = ajAcdGetBoolean("pickanyway");
mispriming_library = ajAcdGetInfile("mispriminglibraryfile");
max_mispriming = ajAcdGetFloat("maxmispriming");
pair_max_mispriming = ajAcdGetFloat("pairmaxmispriming");
gc_clamp = ajAcdGetInt("gcclamp");
opt_size = ajAcdGetInt("osize");
min_size = ajAcdGetInt("minsize");
max_size = ajAcdGetInt("maxsize");
opt_tm = ajAcdGetFloat("otm");
min_tm = ajAcdGetFloat("mintm");
max_tm = ajAcdGetFloat("maxtm");
max_diff_tm = ajAcdGetFloat("maxdifftm");
opt_gc_percent = ajAcdGetFloat("ogcpercent");
min_gc = ajAcdGetFloat("mingc");
max_gc = ajAcdGetFloat("maxgc");
salt_conc = ajAcdGetFloat("saltconc");
dna_conc = ajAcdGetFloat("dnaconc");
num_ns_accepted = ajAcdGetInt("numnsaccepted");
self_any = ajAcdGetFloat("selfany");
self_end = ajAcdGetFloat("selfend");
max_poly_x = ajAcdGetInt("maxpolyx");
AJCNEW0(pipeto,2);
AJCNEW0(pipefrom,2);
/* Sequence Quality */
/* these are not (yet) implemented */
/*
sequence_quality = ajAcdGetInfile("sequencequality");
min_quality = ajAcdGetInt("minquality");
min_end_quality = ajAcdGetInt("minendquality");
quality_range_min = ajAcdGetInt("qualityrangemin");
quality_range_max = ajAcdGetInt("qualityrangemax");
*/
/* Product details */
product_opt_size = ajAcdGetInt("psizeopt");
product_size_range = ajAcdGetRange("prange");
product_opt_tm = ajAcdGetFloat("ptmopt");
product_min_tm = ajAcdGetFloat("ptmmin");
product_max_tm = ajAcdGetFloat("ptmmax");
/* Objective Function Penalty Weights for Primers */
max_end_stability = ajAcdGetFloat("maxendstability");
/* these are not (yet) implemented */
/*
inside_penalty = ajAcdGetFloat("insidepenalty");
outside_penalty = ajAcdGetFloat("outsidepenalty");
*/
/* Primer penalties */
/* these are not (yet) implemented */
/* Internal Oligo "Sequence" Input Tags */
internal_oligo_excluded_region = ajAcdGetRange("oexcludedregion");
internal_oligo_input = ajAcdGetString("oligoinput");
/* Internal Oligo "Global" Input Tags */
internal_oligo_opt_size = ajAcdGetInt("osizeopt");
internal_oligo_min_size = ajAcdGetInt("ominsize");
internal_oligo_max_size = ajAcdGetInt("omaxsize");
internal_oligo_opt_tm = ajAcdGetFloat("otmopt");
internal_oligo_min_tm = ajAcdGetFloat("otmmin");
internal_oligo_max_tm = ajAcdGetFloat("otmmax");
internal_oligo_opt_gc_percent = ajAcdGetFloat("ogcopt");
internal_oligo_min_gc = ajAcdGetFloat("ogcmin");
internal_oligo_max_gc = ajAcdGetFloat("ogcmax");
internal_oligo_salt_conc = ajAcdGetFloat("osaltconc");
internal_oligo_dna_conc = ajAcdGetFloat("odnaconc");
internal_oligo_self_any = ajAcdGetFloat("oanyself");
internal_oligo_self_end = ajAcdGetFloat("oendself");
internal_oligo_max_poly_x = ajAcdGetInt("opolyxmax");
internal_oligo_mishyb_library = ajAcdGetInfile("mishyblibraryfile");
internal_oligo_max_mishyb = ajAcdGetFloat("omishybmax");
/*
internal_oligo_min_quality = ajAcdGetInt("oligominquality");
*/
/* Internal Oligo penalties */
/* these are not (yet) implemented */
/* EMBOSS-wrapper-specific stuff */
outfile = ajAcdGetOutfile("outfile");
ajStrRemoveWhite(&left_input);
ajStrRemoveWhite(&right_input);
/*
** OK - we will now try to do a separate fork-exec for each sequence.
*/
result = ajStrNew();
while(ajSeqallNext(sequence, &seq))
{
program = ajAcdGetpathC("primer3_core");
if(!ajSysExecRedirectC(ajStrGetPtr(program),&pipeto,&pipefrom))
ajFatal("eprimer3: Could not exec primer3_core");
stream = eprimer3_start_write(pipeto[1]);
/* send primer3 Primer "Global" parameters */
eprimer3_send_bool(stream, "PRIMER_EXPLAIN_FLAG", explain_flag);
eprimer3_send_bool(stream, "PRIMER_FILE_FLAG", file_flag);
if(do_hybrid)
{
if(!ajStrCmpC(task[0], "1"))
ajStrAssignC(&taskstr, "pick_pcr_primers_and_hyb_probe");
else if(!ajStrCmpC(task[0], "2"))
ajStrAssignC(&taskstr, "pick_left_only");
else if(!ajStrCmpC(task[0], "3"))
ajStrAssignC(&taskstr, "pick_right_only");
else if(!ajStrCmpC(task[0], "4"))
ajStrAssignC(&taskstr, "pick_hyb_probe_only");
if (!do_primer)
ajStrAssignC(&taskstr, "pick_hyb_probe_only");
}
else
{
if(!ajStrCmpC(task[0], "1"))
ajStrAssignC(&taskstr, "pick_pcr_primers");
else if(!ajStrCmpC(task[0], "2"))
ajStrAssignC(&taskstr, "pick_left_only");
else if(!ajStrCmpC(task[0], "3"))
ajStrAssignC(&taskstr, "pick_right_only");
else if(!ajStrCmpC(task[0], "4"))
ajStrAssignC(&taskstr, "pick_hyb_probe_only");
}
eprimer3_send_string(stream, "PRIMER_TASK", taskstr);
eprimer3_send_int(stream, "PRIMER_NUM_RETURN", num_return);
eprimer3_send_int(stream, "PRIMER_FIRST_BASE_INDEX",
first_base_index);
eprimer3_send_bool(stream, "PRIMER_PICK_ANYWAY", pick_anyway);
/* mispriming library may not have been specified */
if(mispriming_library)
eprimer3_send_stringC(stream, "PRIMER_MISPRIMING_LIBRARY",
ajFileGetPrintnameC(mispriming_library));
eprimer3_send_float(stream, "PRIMER_MAX_MISPRIMING",
max_mispriming);
eprimer3_send_float(stream, "PRIMER_PAIR_MAX_MISPRIMING",
pair_max_mispriming);
eprimer3_send_int(stream, "PRIMER_GC_CLAMP", gc_clamp);
eprimer3_send_int(stream, "PRIMER_OPT_SIZE", opt_size);
eprimer3_send_int(stream, "PRIMER_MIN_SIZE", min_size);
eprimer3_send_int(stream, "PRIMER_MAX_SIZE", max_size);
eprimer3_send_float(stream, "PRIMER_OPT_TM", opt_tm);
eprimer3_send_float(stream, "PRIMER_MIN_TM", min_tm);
eprimer3_send_float(stream, "PRIMER_MAX_TM", max_tm);
eprimer3_send_float(stream, "PRIMER_MAX_DIFF_TM", max_diff_tm);
eprimer3_send_float(stream, "PRIMER_OPT_GC_PERCENT",
opt_gc_percent);
eprimer3_send_float(stream, "PRIMER_MIN_GC", min_gc);
eprimer3_send_float(stream, "PRIMER_MAX_GC", max_gc);
eprimer3_send_float(stream, "PRIMER_SALT_CONC", salt_conc);
eprimer3_send_float(stream, "PRIMER_DNA_CONC", dna_conc);
eprimer3_send_int(stream, "PRIMER_NUM_NS_ACCEPTED",
num_ns_accepted);
eprimer3_send_float(stream, "PRIMER_SELF_ANY", self_any);
eprimer3_send_float(stream, "PRIMER_SELF_END", self_end);
eprimer3_send_int(stream, "PRIMER_MAX_POLY_X", max_poly_x);
eprimer3_send_int(stream, "PRIMER_PRODUCT_OPT_SIZE",
product_opt_size);
eprimer3_send_range2(stream, "PRIMER_PRODUCT_SIZE_RANGE",
product_size_range);
eprimer3_send_float(stream, "PRIMER_PRODUCT_OPT_TM",
product_opt_tm);
eprimer3_send_float(stream, "PRIMER_PRODUCT_MIN_TM",
product_min_tm);
eprimer3_send_float(stream, "PRIMER_PRODUCT_MAX_TM",
product_max_tm);
eprimer3_send_float(stream, "PRIMER_MAX_END_STABILITY",
max_end_stability);
/* send primer3 Internal Oligo "Global" parameters */
eprimer3_send_int(stream, "PRIMER_INTERNAL_OLIGO_OPT_SIZE",
internal_oligo_opt_size);
eprimer3_send_int(stream, "PRIMER_INTERNAL_OLIGO_MIN_SIZE",
internal_oligo_min_size);
eprimer3_send_int(stream, "PRIMER_INTERNAL_OLIGO_MAX_SIZE",
internal_oligo_max_size);
eprimer3_send_float(stream, "PRIMER_INTERNAL_OLIGO_OPT_TM",
internal_oligo_opt_tm);
eprimer3_send_float(stream, "PRIMER_INTERNAL_OLIGO_MIN_TM",
internal_oligo_min_tm);
eprimer3_send_float(stream, "PRIMER_INTERNAL_OLIGO_MAX_TM",
internal_oligo_max_tm);
eprimer3_send_float(stream, "PRIMER_INTERNAL_OLIGO_OPT_GC_PERCENT",
internal_oligo_opt_gc_percent);
eprimer3_send_float(stream, "PRIMER_INTERNAL_OLIGO_MIN_GC",
internal_oligo_min_gc);
eprimer3_send_float(stream, "PRIMER_INTERNAL_OLIGO_MAX_GC",
internal_oligo_max_gc);
eprimer3_send_float(stream, "PRIMER_INTERNAL_OLIGO_SALT_CONC",
internal_oligo_salt_conc);
eprimer3_send_float(stream, "PRIMER_INTERNAL_OLIGO_DNA_CONC",
internal_oligo_dna_conc);
eprimer3_send_float(stream, "PRIMER_INTERNAL_OLIGO_SELF_ANY",
internal_oligo_self_any);
eprimer3_send_float(stream, "PRIMER_INTERNAL_OLIGO_SELF_END",
internal_oligo_self_end);
eprimer3_send_int(stream, "PRIMER_INTERNAL_OLIGO_MAX_POLY_X",
internal_oligo_max_poly_x);
/*
** internal oligo mishybridising library may not have been
** specified
*/
if(internal_oligo_mishyb_library)
eprimer3_send_stringC(stream,
"PRIMER_INTERNAL_OLIGO_MISHYB_LIBRARY",
ajFileGetPrintnameC(internal_oligo_mishyb_library));
eprimer3_send_float(stream, "PRIMER_INTERNAL_OLIGO_MAX_MISHYB",
internal_oligo_max_mishyb);
/*
** Start sequence-specific stuff
*/
begin = ajSeqallGetseqBegin(sequence) - 1;
end = ajSeqallGetseqEnd(sequence) - 1;
strand = ajSeqGetSeqCopyS(seq);
ajStrFmtUpper(&strand);
ajStrAssignSubC(&substr,ajStrGetPtr(strand), begin, end);
/* send flags to turn on using optimal product size */
eprimer3_send_float(stream, "PRIMER_PAIR_WT_PRODUCT_SIZE_GT",
(float)0.05);
eprimer3_send_float(stream, "PRIMER_PAIR_WT_PRODUCT_SIZE_LT",
(float)0.05);
/* send primer3 Primer "Sequence" parameters */
eprimer3_send_string(stream, "SEQUENCE", substr);
/* if no ID name, use the USA */
if(ajStrMatchC(ajSeqGetNameS(seq),""))
eprimer3_send_string(stream, "PRIMER_SEQUENCE_ID",
ajSeqGetUsaS(seq));
else
eprimer3_send_string(stream, "PRIMER_SEQUENCE_ID",
ajSeqGetNameS(seq));
eprimer3_send_range(stream, "INCLUDED_REGION", included_region,
begin);
eprimer3_send_range(stream, "TARGET", target, begin);
eprimer3_send_range(stream, "EXCLUDED_REGION", excluded_region,
begin);
eprimer3_send_string(stream, "PRIMER_LEFT_INPUT", left_input);
eprimer3_send_string(stream, "PRIMER_RIGHT_INPUT", right_input);
/* send primer3 Internal Oligo "Sequence" parameters */
eprimer3_send_range(stream,
"PRIMER_INTERNAL_OLIGO_EXCLUDED_REGION",
internal_oligo_excluded_region, begin);
eprimer3_send_string(stream, "PRIMER_INTERNAL_OLIGO_INPUT",
internal_oligo_input);
/* end the primer3 input sequence record with a '=' */
eprimer3_send_end(stream);
/* and close the ouput pipe stream */
eprimer3_end_write(stream);
/* read the primer3 output */
eprimer3_read(pipefrom[0], &result);
eprimer3_report(outfile, result, num_return, begin);
ajStrSetClear(&result);
#ifndef WIN32
close(pipeto[1]);
close(pipefrom[0]);
#endif
} /* end of sequence loop */
ajStrDel(&result);
ajSeqDel(&seq);
ajStrDel(&strand);
ajStrDel(&substr);
ajFileClose(&outfile);
ajStrDel(&taskstr);
ajStrDelarray(&task);
ajSeqallDel(&sequence);
ajSeqDel(&seq);
ajRangeDel(&included_region);
ajRangeDel(&target);
ajRangeDel(&excluded_region);
ajRangeDel(&product_size_range);
ajRangeDel(&internal_oligo_excluded_region);
ajStrDel(&left_input);
ajStrDel(&right_input);
ajStrDel(&internal_oligo_input);
AJFREE(pipeto);
AJFREE(pipefrom);
ajFileClose(&mispriming_library);
embExit();
return 0;
}
EmbPMatPrints embMatProtReadInt(AjPFile fp)
{
EmbPMatPrints ret;
AjPStr line;
ajint i;
ajuint j;
ajuint m;
const char *p;
line = ajStrNewC("#");
p = ajStrGetPtr(line);
while(!*p || *p=='#' || *p=='!' || *p=='\n')
{
if(!ajReadlineTrim(fp,&line))
{
ajStrDel(&line);
return NULL;
}
p = ajStrGetPtr(line);
}
ajDebug("embMatProtReadint starting\n");
ajDebug ("Line: %S\n", line);
AJNEW0 (ret);
ret->cod = ajStrNew();
ajStrAssignS(&ret->cod,line);
ajReadlineTrim(fp,&line);
ret->acc = ajStrNew();
ajStrAssignS(&ret->acc,line);
ajReadlineTrim(fp,&line);
ajStrToUint(line,&ret->n);
ajReadlineTrim(fp,&line);
ret->tit = ajStrNew();
ajStrAssignS(&ret->tit,line);
ajDebug ("Lineb: %S\n", line);
AJCNEW(ret->len, ret->n);
AJCNEW(ret->max, ret->n);
AJCNEW(ret->thresh, ret->n);
AJCNEW(ret->matrix, ret->n);
for(m=0;m<ret->n;++m)
{
ajReadlineTrim(fp,&line);
ajStrToUint(line,&ret->len[m]);
ajReadlineTrim(fp,&line);
ajStrToUint(line,&ret->thresh[m]);
ajReadlineTrim(fp,&line);
ajStrToUint(line,&ret->max[m]);
ajDebug ("m: %d/%d len:%d thresh:%d max:%d\n",
m, ret->n, ret->len[m], ret->thresh[m], ret->max[m]);
for(i=0;i<26;++i)
{
AJCNEW0(ret->matrix[m][i], ret->len[m]);
ajReadlineTrim(fp,&line);
ajDebug ("Linec [%d][%d]: %S\n", m, i, line);
p = ajStrGetPtr(line);
for(j=0;j<ret->len[m];++j)
{
if(!j)
p = ajSysFuncStrtok(p," ");
else
p = ajSysFuncStrtok(NULL," ");
sscanf(p,"%u",&ret->matrix[m][i][j]);
}
}
}
ajReadlineTrim(fp,&line);
ajDebug ("Linec: %S\n", line);
ajStrDel(&line);
return ret;
}
int main(int argc, char **argv)
{
AjPFile outf = NULL;
AjPSeq sequence = NULL;
AjPStr substr = NULL;
AjPStr seqstr = NULL;
AjPStr revstr = NULL;
AjPStr p1;
AjPStr p2;
PPrimer eric = NULL;
PPrimer fred = NULL;
PPrimer f;
PPrimer r;
PPair pair;
AjPList forlist = NULL;
AjPList revlist = NULL;
AjPList pairlist = NULL;
AjBool targetrange;
AjBool isDNA = ajTrue;
AjBool dolist = ajFalse;
ajint primerlen = 0;
ajint minprimerlen = 0;
ajint maxprimerlen = 0;
ajint minprodlen = 0;
ajint maxprodlen = 0;
ajint prodlen = 0;
ajint seqlen = 0;
ajint stepping_value = 1;
ajint targetstart = 0;
ajint targetend = 0;
ajint limit = 0;
ajint limit2 = 0;
ajint lastpos = 0;
ajint startpos = 0;
ajint endpos = 0;
ajint begin;
ajint end;
ajint v1;
ajint v2;
ajint overlap;
float minpmGCcont = 0.;
float maxpmGCcont = 0.;
float minprodGCcont = 0.;
float maxprodGCcont = 0.;
float prodTm;
float prodGC;
ajint i;
ajint j;
ajint neric=0;
ajint nfred=0;
ajint npair=0;
float minprimerTm = 0.0;
float maxprimerTm = 0.0;
float saltconc = 0.0;
float dnaconc = 0.0;
embInit ("prima", argc, argv);
substr = ajStrNew();
forlist = ajListNew();
revlist = ajListNew();
pairlist = ajListNew();
p1 = ajStrNew();
p2 = ajStrNew();
sequence = ajAcdGetSeq("sequence");
outf = ajAcdGetOutfile("outfile");
minprimerlen = ajAcdGetInt("minprimerlen");
maxprimerlen = ajAcdGetInt("maxprimerlen");
minpmGCcont = ajAcdGetFloat("minpmGCcont");
maxpmGCcont = ajAcdGetFloat("maxpmGCcont");
minprimerTm = ajAcdGetFloat("mintmprimer");
maxprimerTm = ajAcdGetFloat("maxtmprimer");
minprodlen = ajAcdGetInt("minplen");
maxprodlen = ajAcdGetInt("maxplen");
minprodGCcont = ajAcdGetFloat("minpgccont");
maxprodGCcont = ajAcdGetFloat("maxpgccont");
saltconc = ajAcdGetFloat("saltconc");
dnaconc = ajAcdGetFloat("dnaconc");
targetrange = ajAcdGetToggle("targetrange");
targetstart = ajAcdGetInt("targetstart");
targetend = ajAcdGetInt("targetend");
overlap = ajAcdGetInt("overlap");
dolist = ajAcdGetBoolean("list");
seqstr = ajSeqGetSeqCopyS(sequence);
ajStrFmtUpper(&seqstr);
begin = ajSeqGetBegin(sequence);
end = ajSeqGetEnd(sequence);
seqlen = end-begin+1;
ajStrAssignSubC(&substr,ajStrGetPtr(seqstr),begin-1,end-1);
revstr = ajStrNewC(ajStrGetPtr(substr));
ajSeqstrReverse(&revstr);
AJCNEW0(entropy, seqlen);
AJCNEW0(enthalpy, seqlen);
AJCNEW0(energy, seqlen);
/* Initialise Tm calculation arrays */
ajMeltTempSave(ajStrGetPtr(substr),0,seqlen,saltconc,dnaconc,1,
&entropy, &enthalpy, &energy);
ajFmtPrintF(outf, "\n\nINPUT SUMMARY\n");
ajFmtPrintF(outf, "*************\n\n");
if(targetrange)
ajFmtPrintF
(outf, "Prima of %s from positions %d to %d bps\n",
ajSeqGetNameC(sequence),targetstart, targetend);
else
ajFmtPrintF(outf, "Prima of %s\n", ajSeqGetNameC(sequence));
ajFmtPrintF(outf, "PRIMER CONSTRAINTS:\n");
ajFmtPrintF
(outf, "PRIMA DOES NOT ALLOW PRIMER SEQUENCE AMBIGUITY OR ");
ajFmtPrintF(outf,"DUPLICATE PRIMER ENDPOINTS\n");
ajFmtPrintF(outf,
"Primer size range is %d-%d\n",minprimerlen,maxprimerlen);
ajFmtPrintF(outf,
"Primer GC content range is %.2f-%.2f\n",minpmGCcont,
maxpmGCcont);
ajFmtPrintF(outf,"Primer melting Temp range is %.2f - %.2f C\n",
minprimerTm, maxprimerTm);
ajFmtPrintF (outf, "PRODUCT CONSTRAINTS:\n");
ajFmtPrintF(outf,"Product GC content range is %.2f-%.2f\n",
minprodGCcont, maxprodGCcont);
ajFmtPrintF(outf, "Salt concentration is %.2f (mM)\n", saltconc);
ajFmtPrintF(outf, "DNA concentration is %.2f (nM)\n", dnaconc);
if(targetrange)
ajFmtPrintF(outf, "Targeted range to amplify is from %d to %d\n",
targetstart,targetend);
else
{
ajFmtPrintF(outf,"Considering all suitable Primer pairs with ");
ajFmtPrintF(outf,"Product length ranges %d to %d\n\n\n", minprodlen,
maxprodlen);
}
ajFmtPrintF(outf, "\n\nPRIMER/PRODUCT PAIR CALCULATIONS & OUTPUT\n");
ajFmtPrintF(outf, "*****************************************\n\n");
if(seqlen-minprimerlen < 0)
ajFatal("Sequence too short");
if(targetrange)
{
ajStrAssignSubC(&p1,ajStrGetPtr(substr),targetstart-begin,targetend-begin);
prodGC = ajMeltGC(substr,seqlen);
prodTm = ajMeltTempProd(prodGC,saltconc,seqlen);
if(prodGC<minprodGCcont || prodGC>maxprodGCcont)
{
ajFmtPrintF(outf,
"Product GC content [%.2f] outside acceptable range\n",
prodGC);
embExitBad();
return 0;
}
prima_testtarget(substr, revstr, targetstart-begin, targetend-begin,
minprimerlen, maxprimerlen,
seqlen, minprimerTm, maxprimerTm, minpmGCcont,
maxpmGCcont, minprodGCcont, maxprodGCcont, saltconc,
dnaconc, pairlist, &npair);
}
if(!targetrange)
{
limit = seqlen-minprimerlen-minprodlen+1;
lastpos = seqlen-minprodlen;
limit2 = maxprodlen-minprodlen;
/* Outer loop selects all possible product start points */
for(i=minprimerlen; i<limit; ++i)
{
startpos = i;
ajDebug("Position in sequence %d\n",startpos);
endpos = i+minprodlen-1;
/* Inner loop selects all possible product lengths */
for(j=0; j<limit2; ++j, ++endpos)
{
if(endpos>lastpos)
break;
v1 = endpos-startpos+1;
ajStrAssignSubC(&p1,ajStrGetPtr(substr),startpos,endpos);
prodGC = ajMeltGC(p1,v1);
prodTm = ajMeltTempProd(prodGC,saltconc,v1);
if(prodGC<minprodGCcont || prodGC>maxprodGCcont)
continue;
/* Only accept primers with acceptable Tm and GC */
neric = 0;
nfred = 0;
prima_testproduct(substr, startpos, endpos, primerlen,
minprimerlen, maxprimerlen,minpmGCcont,
maxpmGCcont, minprimerTm, maxprimerTm,
minprodlen, maxprodlen, prodTm, prodGC, seqlen,
&eric,&fred,forlist,revlist,&neric,&nfred,
stepping_value, saltconc,dnaconc, isDNA, begin);
if(!neric)
continue;
/* Now reject those primers with self-complementarity */
prima_reject_self(forlist,revlist,&neric,&nfred);
if(!neric)
continue;
/* Reject any primers that could bind elsewhere in the
sequence */
prima_test_multi(forlist,revlist,&neric,&nfred,substr,revstr,
seqlen);
/* Now select the least complementary pair (if any) */
prima_best_primer(forlist, revlist, &neric, &nfred);
if(!neric)
continue;
AJNEW(pair);
ajListPop(forlist,(void **)&f);
ajListPop(revlist,(void **)&r);
pair->f = f;
pair->r = r;
++npair;
ajListPush(pairlist,(void *)pair);
}
}
}
if(!targetrange)
{
/* Get rid of primer pairs nearby the top scoring ones */
prima_TwoSortscorepos(&pairlist);
prima_prune_nearby(pairlist, &npair, maxprimerlen-1);
ajListSort(pairlist,prima_PosCompare);
prima_check_overlap(pairlist,&npair,overlap);
}
if(npair)
{
if(!targetrange)
ajFmtPrintF(outf,"%d pairs found\n\n",npair);
else
ajFmtPrintF(outf,
"Closest primer pair to specified product is:\n\n");
if((maxprimerlen<26 && seqlen<999999 && !dolist))
ajFmtPrintF(outf,"\n\t\tForward\t\t\t\t\tReverse\n\n");
}
for(i=0;i<npair;++i)
{
if(!targetrange)
ajFmtPrintF(outf,"[%d]\n",i+1);
ajListPop(pairlist,(void **)&pair);
prodlen = pair->r->start - (pair->f->start + pair->f->primerlen);
if((maxprimerlen<26 && seqlen<999999 && !dolist))
{
v1 = pair->f->start;
v2 = v1 + pair->f->primerlen -1;
ajStrAssignSubS(&p1,substr,v1,v2);
ajFmtPrintF(outf,"%6d %-25.25s %d\t", v1+begin, ajStrGetPtr(p1),
v2+begin);
v1 = pair->r->start;
v2 = v1 + pair->r->primerlen -1;
ajStrAssignSubS(&p2,substr,v1,v2);
ajSeqstrReverse(&p2);
ajFmtPrintF(outf,
"%6d %-25.25s %d\n", v1+begin, ajStrGetPtr(p2), v2+begin);
ajFmtPrintF(outf," Tm %.2f C (GC %.2f%%)\t\t ",
pair->f->primerTm,pair->f->primGCcont*100.);
ajFmtPrintF(outf,"Tm %.2f C (GC %.2f%%)\n",
pair->r->primerTm,pair->r->primGCcont*100.);
ajFmtPrintF(outf," Length: %-32dLength: %d\n",
pair->f->primerlen,pair->r->primerlen);
ajFmtPrintF(outf," Tma: %.2f C\t\t\t",
ajAnneal(pair->f->primerTm,pair->f->prodTm));
ajFmtPrintF(outf," Tma: %.2f C\n\n\n",
ajAnneal(pair->r->primerTm,pair->f->prodTm));
ajFmtPrintF(outf," Product GC: %.2f%%\n",
pair->f->prodGC * 100.0);
ajFmtPrintF(outf," Product Tm: %.2f C\n",
pair->f->prodTm);
ajFmtPrintF(outf," Length: %d\n\n\n",prodlen);
}
else
{
ajFmtPrintF(outf," Product from %d to %d\n",pair->f->start+
pair->f->primerlen+begin,pair->r->start-1+begin);
ajFmtPrintF(outf," Tm: %.2f C GC: %.2f%%\n",
pair->f->prodTm,pair->f->prodGC*(float)100.);
ajFmtPrintF(outf," Length: %d\n\n\n",prodlen);
v1 = pair->f->start;
v2 = v1 + pair->f->primerlen -1;
ajStrAssignSubS(&p1,substr,v1,v2);
ajFmtPrintF(outf," Forward: 5' %s 3'\n",ajStrGetPtr(p1));
ajFmtPrintF(outf," Start: %d\n",v1+begin);
ajFmtPrintF(outf," End: %d\n",v2+begin);
ajFmtPrintF(outf," Tm: %.2f C\n",
pair->f->primerTm);
ajFmtPrintF(outf," GC: %.2f%%\n",
pair->f->primGCcont*(float)100.);
ajFmtPrintF(outf," Len: %d\n",
pair->f->primerlen);
ajFmtPrintF(outf," Tma: %.2f C\n\n\n",
ajAnneal(pair->f->primerTm,pair->f->prodTm));
v1 = pair->r->start;
v2 = v1 + pair->r->primerlen -1;
ajStrAssignSubS(&p2,substr,v1,v2);
ajSeqstrReverse(&p2);
ajStrAssignSubS(&p1,substr,v1,v2);
ajFmtPrintF(outf," Reverse: 5' %s 3'\n",ajStrGetPtr(p1));
ajFmtPrintF(outf," Start: %d\n",v1+begin);
ajFmtPrintF(outf," End: %d\n",v2+begin);
ajFmtPrintF(outf," Tm: %.2f C\n",
pair->r->primerTm);
ajFmtPrintF(outf," GC: %.2f%%\n",
pair->r->primGCcont*(float)100.);
ajFmtPrintF(outf," Len: %d\n",
pair->r->primerlen);
ajFmtPrintF(outf," Tma: %.2f C\n\n\n",
ajAnneal(pair->r->primerTm,pair->f->prodTm));
}
prima_PrimerDel(&pair->f);
prima_PrimerDel(&pair->r);
AJFREE(pair);
}
ajStrDel(&seqstr);
ajStrDel(&revstr);
ajStrDel(&substr);
ajStrDel(&p1);
ajStrDel(&p2);
ajListFree(&forlist);
ajListFree(&revlist);
ajListFree(&pairlist);
ajFileClose(&outf);
ajSeqDel(&sequence);
AJFREE(entropy);
AJFREE(enthalpy);
AJFREE(energy);
embExit();
return 0;
}
int main(int argc, char **argv)
{
AjPAlign align;
AjPSeqall seqall;
AjPSeq a;
AjPSeq b;
AjPStr alga;
AjPStr algb;
AjPStr ss;
ajuint lena;
ajuint lenb;
const char *p;
const char *q;
ajint start1 = 0;
ajint start2 = 0;
ajint *compass;
float* ix;
float* iy;
float* m;
AjPMatrixf matrix;
AjPSeqCvt cvt = 0;
float **sub;
float gapopen;
float gapextend;
float endgapopen;
float endgapextend;
size_t maxarr = 1000; /* arbitrary. realloc'd if needed */
size_t len;
float score;
AjBool dobrief = ajTrue;
AjBool endweight = ajFalse; /* should end gap penalties be applied */
float id = 0.;
float sim = 0.;
float idx = 0.;
float simx = 0.;
AjPStr tmpstr = NULL;
embInit("needle", argc, argv);
matrix = ajAcdGetMatrixf("datafile");
a = ajAcdGetSeq("asequence");
ajSeqTrim(a);
seqall = ajAcdGetSeqall("bsequence");
gapopen = ajAcdGetFloat("gapopen");
gapextend = ajAcdGetFloat("gapextend");
endgapopen = ajAcdGetFloat("endopen");
endgapextend = ajAcdGetFloat("endextend");
dobrief = ajAcdGetBoolean("brief");
endweight = ajAcdGetBoolean("endweight");
align = ajAcdGetAlign("outfile");
gapopen = ajRoundFloat(gapopen, 8);
gapextend = ajRoundFloat(gapextend, 8);
AJCNEW0(compass, maxarr);
AJCNEW0(m, maxarr);
AJCNEW0(ix, maxarr);
AJCNEW0(iy, maxarr);
alga = ajStrNew();
algb = ajStrNew();
ss = ajStrNew();
sub = ajMatrixfGetMatrix(matrix);
cvt = ajMatrixfGetCvt(matrix);
lena = ajSeqGetLen(a);
while(ajSeqallNext(seqall,&b))
{
ajSeqTrim(b);
lenb = ajSeqGetLen(b);
if(lenb > (LONG_MAX/(size_t)(lena+1)))
ajDie("Sequences too big. Try 'stretcher'");
len = (size_t)lena*(size_t)lenb;
if(len>maxarr)
{
AJCRESIZETRY0(compass,(size_t)maxarr,len);
if(!compass)
ajDie("Sequences too big. Try 'stretcher'");
AJCRESIZETRY0(m,(size_t)maxarr,len);
if(!m)
ajDie("Sequences too big. Try 'stretcher'");
AJCRESIZETRY0(ix,(size_t)maxarr,len);
if(!ix)
ajDie("Sequences too big. Try 'stretcher'");
AJCRESIZETRY0(iy,(size_t)maxarr,len);
if(!iy)
ajDie("Sequences too big. Try 'stretcher'");
maxarr=len;
}
p = ajSeqGetSeqC(a);
q = ajSeqGetSeqC(b);
ajStrAssignC(&alga,"");
ajStrAssignC(&algb,"");
score = embAlignPathCalcWithEndGapPenalties(p, q, lena, lenb,
gapopen, gapextend, endgapopen, endgapextend,
&start1, &start2, sub, cvt,
m, ix, iy, compass, ajFalse, endweight);
embAlignWalkNWMatrixUsingCompass(p, q, &alga, &algb,
lena, lenb, &start1, &start2,
compass);
embAlignReportGlobal(align, a, b, alga, algb,
start1, start2,
gapopen, gapextend,
score, matrix,
ajSeqGetOffset(a), ajSeqGetOffset(b));
if(!dobrief)
{
embAlignCalcSimilarity(alga,algb,sub,cvt,lena,lenb,&id,&sim,&idx,
&simx);
ajFmtPrintS(&tmpstr,"Longest_Identity = %5.2f%%\n",
id);
ajFmtPrintAppS(&tmpstr,"Longest_Similarity = %5.2f%%\n",
sim);
ajFmtPrintAppS(&tmpstr,"Shortest_Identity = %5.2f%%\n",
idx);
ajFmtPrintAppS(&tmpstr,"Shortest_Similarity = %5.2f%%",
simx);
ajAlignSetSubHeaderApp(align, tmpstr);
}
ajAlignWrite(align);
ajAlignReset(align);
}
ajAlignClose(align);
ajAlignDel(&align);
ajSeqallDel(&seqall);
ajSeqDel(&a);
ajSeqDel(&b);
AJFREE(compass);
AJFREE(ix);
AJFREE(iy);
AJFREE(m);
ajStrDel(&alga);
ajStrDel(&algb);
ajStrDel(&ss);
ajStrDel(&tmpstr);
embExit();
return 0;
}
int main(int argc, char **argv)
{
AjPList idlist;
AjPList* fieldList = NULL;
AjBool systemsort;
AjBool cleanup;
ajuint maxindex;
ajuint maxidlen = 0;
ajuint maxlen;
AjPFile elistfile = NULL;
AjPFile* alistfile = NULL;
AjPStr dbname = NULL;
AjPStr release = NULL;
AjPStr datestr = NULL;
AjPStr sortopt = NULL;
void **entryIds = NULL;
AjPStr directory;
AjPStr indexdir;
AjPStr filename;
AjPStr exclude;
AjPStr curfilename = NULL;
AjPFile libr=NULL;
AjPStr idformat = NULL;
EmbPEntry entry;
ajuint idtype = 0;
ajuint idCount = 0;
ajuint idDone;
AjPList listInputFiles = NULL;
void ** inputFiles = NULL;
ajuint nfiles;
ajuint ifile;
ajuint filesize;
short recsize;
ajuint maxfilelen = 20;
char date[4] =
{
0,0,0,0
};
AjPStr tmpfname = NULL;
AjPStr* fields = NULL;
AjPFile entFile = NULL;
AjPStr* divfiles = NULL;
AjPRegexp regIdExp = NULL;
ajint* maxFieldLen = NULL;
ajuint ifield = 0;
ajuint nfields = 0;
AjPFile logfile = NULL;
ajuint* countField = NULL;
ajuint* fieldTot = NULL;
ajuint idCountFile = 0;
ajuint i;
embInit("dbifasta", argc, argv);
idformat = ajAcdGetListSingle("idformat");
fields = ajAcdGetList("fields");
directory = ajAcdGetDirectoryName("directory");
indexdir = ajAcdGetOutdirName("indexoutdir");
filename = ajAcdGetString("filenames");
exclude = ajAcdGetString("exclude");
dbname = ajAcdGetString("dbname");
release = ajAcdGetString("release");
datestr = ajAcdGetString("date");
systemsort = ajAcdGetBoolean("systemsort");
cleanup = ajAcdGetBoolean("cleanup");
sortopt = ajAcdGetString("sortoptions");
maxindex = ajAcdGetInt("maxindex");
logfile = ajAcdGetOutfile("outfile");
while(fields[nfields]) /* array ends with a NULL */
nfields++;
if(nfields)
{
AJCNEW(maxFieldLen, nfields);
AJCNEW0(countField, nfields);
AJCNEW0(fieldTot, nfields);
for(ifield=0; ifield < nfields; ifield++)
maxFieldLen[ifield] = (ajint)maxindex * -1;
if(systemsort)
AJCNEW(alistfile, nfields);
else
{
AJCNEW(fieldList, nfields);
for(ifield=0; ifield < nfields; ifield++)
fieldList[ifield] = ajListNew();
}
}
if(ajStrMatchC(datestr, "00/00/00"))
ajFmtPrintS(&datestr, "%D", ajTimeRefTodayFmt("dbindex"));
ajStrRemoveWhite(&dbname); /* used for temp filenames */
embDbiDateSet(datestr, date);
idlist = ajListNew();
regIdExp = dbifasta_getExpr(idformat, &idtype);
ajDebug("reading '%S/%S'\n", directory, filename);
ajDebug("writing '%S/'\n", indexdir);
listInputFiles = embDbiFileListExc(directory, filename, exclude);
ajListSort(listInputFiles, &ajStrVcmp);
nfiles = (ajuint) ajListToarray(listInputFiles, &inputFiles);
if(!nfiles)
ajDie("No input files in '%S' matched filename '%S'",
directory, filename);
embDbiLogHeader(logfile, dbname, release, datestr,
indexdir, maxindex);
embDbiLogFields(logfile, fields, nfields);
embDbiLogSource(logfile, directory, filename, exclude,
(AjPStr*) inputFiles, nfiles);
embDbiLogCmdline(logfile);
AJCNEW0(divfiles, nfiles);
/*
** process each input file, one at a time
*/
for(ifile=0; ifile < nfiles; ifile++)
{
ajStrAssignS(&curfilename,(AjPStr) inputFiles[ifile]);
embDbiFlatOpenlib(curfilename, &libr);
ajFilenameTrimPath(&curfilename);
if(ajStrGetLen(curfilename) >= maxfilelen)
maxfilelen = ajStrGetLen(curfilename) + 1;
ajDebug("processing filename '%S' ...\n", curfilename);
ajDebug("processing file '%F' ...\n", libr);
ajStrAssignS(&divfiles[ifile], curfilename);
if(systemsort) /* elistfile for entries, alist for fields */
elistfile = embDbiSortOpen(alistfile, ifile,
dbname, fields, nfields);
idCountFile = 0;
for(i=0;i<nfields;i++)
countField[i] = 0;
while((entry=dbifasta_NextFlatEntry(libr, ifile,
regIdExp, idtype,
systemsort, fields,
maxFieldLen, &maxidlen,
countField, elistfile,
alistfile)))
{
idCountFile++;
if(!systemsort) /* save the entry data in lists */
embDbiMemEntry(idlist, fieldList, nfields,
entry, ifile);
entry = NULL;
}
idCount += idCountFile;
if(systemsort)
{
embDbiSortClose(&elistfile, alistfile, nfields);
AJFREE(entry);
}
else
{
embDbiEntryDel(&dbifastaGEntry);
}
embDbiLogFile(logfile, curfilename, idCountFile, fields,
countField, nfields);
}
/* write the division.lkp file */
embDbiWriteDivision(indexdir, dbname, release, date,
maxfilelen, nfiles, divfiles, NULL);
/* Write the entryname.idx index */
ajStrAssignC(&tmpfname, "entrynam.idx");
entFile = ajFileNewOutNamePathS(tmpfname, indexdir);
recsize = maxidlen+10;
filesize = 300 + (idCount*(ajint)recsize);
embDbiHeader(entFile, filesize, idCount, recsize, dbname, release, date);
if(systemsort)
idDone = embDbiSortWriteEntry(entFile, maxidlen,
dbname, nfiles, cleanup, sortopt);
else /* save entries in entryIds array */
{
idDone = embDbiMemWriteEntry(entFile, maxidlen,
idlist, &entryIds);
if(idDone != idCount)
ajFatal("Duplicates not allowed for in-memory processing");
}
embDbiHeaderSize(entFile, 300+(idDone*(ajint)recsize), idDone);
ajFileClose(&entFile);
/* Write the fields index files */
for(ifield=0; ifield < nfields; ifield++)
{
if(maxindex)
maxlen = maxindex;
else
{
if(maxFieldLen[ifield] >= 0)
maxlen = maxFieldLen[ifield];
else
maxlen = - maxFieldLen[ifield];
}
if(systemsort)
fieldTot[ifield] = embDbiSortWriteFields(dbname, release,
date, indexdir,
fields[ifield], maxlen,
nfiles, idCount,
cleanup, sortopt);
else
fieldTot[ifield] = embDbiMemWriteFields(dbname, release,
date, indexdir,
fields[ifield], maxlen,
fieldList[ifield],
entryIds);
}
embDbiLogFinal(logfile,maxindex, maxFieldLen, fields, fieldTot,
nfields, nfiles, idDone, idCount);
if(systemsort)
embDbiRmEntryFile(dbname, cleanup);
ajStrDel(&idformat);
ajStrDelarray(&fields);
ajStrDel(&filename);
ajStrDel(&exclude);
ajStrDel(&dbname);
ajStrDel(&release);
ajStrDel(&datestr);
ajStrDel(&sortopt);
ajStrDel(&directory);
ajStrDel(&indexdir);
ajStrDel(&tmpfname);
ajFileClose(&libr);
ajFileClose(&logfile);
for(i=0;i<nfields;i++)
{
if(systemsort)
{
ajFileClose(&alistfile[i]);
}
else
{
ajListMap(fieldList[i], &embDbiFieldDelMap, NULL);
ajListFree(&fieldList[i]);
}
}
AJFREE(alistfile);
AJFREE(fieldList);
AJFREE(maxFieldLen);
AJFREE(countField);
AJFREE(fieldTot);
for(i=0;i<nfiles;i++)
{
ajStrDel(&divfiles[i]);
}
AJFREE(divfiles);
AJFREE(inputFiles);
embDbiEntryDel(&dbifastaGEntry);
ajStrDel(&dbifastaGRline);
ajStrDel(&dbifastaGTmpId);
if(dbifastaGFdl)
{
for(i=0; i < nfields; i++)
ajListFree(&dbifastaGFdl[i]);
AJFREE(dbifastaGFdl);
}
ajListMap(idlist, &embDbiEntryDelMap, NULL);
ajListFree(&idlist);
ajListstrFreeData(&listInputFiles);
AJFREE(entryIds);
ajRegFree(&dbifastaGIdexp);
ajRegFree(&dbifastaGWrdexp);
ajRegFree(®IdExp);
ajStrDel(&dbifastaGTmpAc);
ajStrDel(&dbifastaGTmpSv);
ajStrDel(&dbifastaGTmpGi);
ajStrDel(&dbifastaGTmpDb);
ajStrDel(&dbifastaGTmpDes);
ajStrDel(&dbifastaGTmpFd);
ajStrDel(&curfilename);
embExit();
return 0;
}
int main(int argc, char **argv)
{
AjPSeqset seqset;
AjPSeqall seqall;
AjPSeq queryseq;
const AjPSeq targetseq;
ajint wordlen;
AjPTable wordsTable = NULL;
AjPList* matchlist = NULL;
AjPFile logfile;
AjPFeattable* seqsetftables = NULL;
AjPFeattable seqallseqftable = NULL;
AjPFeattabOut ftoutforseqsetseq = NULL;
AjPFeattabOut ftoutforseqallseq = NULL;
AjPAlign align = NULL;
AjIList iter = NULL;
ajint targetstart;
ajint querystart;
ajint len;
ajuint i, j;
ajulong nAllMatches = 0;
ajulong sumAllScore = 0;
AjBool dumpAlign = ajTrue;
AjBool dumpFeature = ajTrue;
AjBool checkmode = ajFalse;
EmbPWordRK* wordsw = NULL;
ajuint npatterns = 0;
ajuint seqsetsize;
ajuint nmatches;
ajuint* nmatchesseqset;
ajuint* lastlocation; /* Cursors for Rabin-Karp search. */
/* Shows until what point the query sequence was
* scanned for a pattern sequences in the seqset.
*/
char* paddedheader = NULL;
const char* header;
AjPStr padding;
header = "Pattern %S #pat-sequences #all-matches avg-match-length\n";
padding = ajStrNew();
embInit("wordmatch", argc, argv);
wordlen = ajAcdGetInt("wordsize");
seqset = ajAcdGetSeqset("asequence");
seqall = ajAcdGetSeqall("bsequence");
logfile = ajAcdGetOutfile("logfile");
dumpAlign = ajAcdGetToggle("dumpalign");
dumpFeature = ajAcdGetToggle("dumpfeat");
if(dumpAlign)
{
align = ajAcdGetAlign("outfile");
ajAlignSetExternal(align, ajTrue);
}
seqsetsize = ajSeqsetGetSize(seqset);
ajSeqsetTrim(seqset);
AJCNEW0(matchlist, seqsetsize);
AJCNEW0(seqsetftables, seqsetsize);
AJCNEW0(nmatchesseqset, seqsetsize);
if (dumpFeature)
{
ftoutforseqsetseq = ajAcdGetFeatout("aoutfeat");
ftoutforseqallseq = ajAcdGetFeatout("boutfeat");
}
checkmode = !dumpFeature && !dumpAlign;
embWordLength(wordlen);
ajFmtPrintF(logfile, "Small sequence/file for constructing"
" target patterns: %S\n", ajSeqsetGetUsa(seqset));
ajFmtPrintF(logfile, "Large sequence/file to be scanned"
" for patterns: %S\n", ajSeqallGetUsa(seqall));
ajFmtPrintF(logfile, "Number of sequences in the patterns file: %u\n",
seqsetsize);
ajFmtPrintF(logfile, "Pattern/word length: %u\n", wordlen);
for(i=0;i<seqsetsize;i++)
{
targetseq = ajSeqsetGetseqSeq(seqset, i);
embWordGetTable(&wordsTable, targetseq);
}
AJCNEW0(lastlocation, seqsetsize);
if(ajTableGetLength(wordsTable)>0)
{
npatterns = embWordRabinKarpInit(wordsTable,
&wordsw, wordlen, seqset);
ajFmtPrintF(logfile, "Number of patterns/words found: %u\n", npatterns);
while(ajSeqallNext(seqall,&queryseq))
{
for(i=0;i<seqsetsize;i++)
{
lastlocation[i]=0;
if (!checkmode)
matchlist[i] = ajListstrNew();
}
nmatches = embWordRabinKarpSearch(
ajSeqGetSeqS(queryseq), seqset,
(EmbPWordRK const *)wordsw, wordlen, npatterns,
matchlist, lastlocation, checkmode);
nAllMatches += nmatches;
if (checkmode)
continue;
for(i=0;i<seqsetsize;i++)
{
if(ajListGetLength(matchlist[i])>0)
{
iter = ajListIterNewread(matchlist[i]) ;
while(embWordMatchIter(iter, &targetstart, &querystart, &len,
&targetseq))
{
if(dumpAlign)
{
ajAlignDefineSS(align, targetseq, queryseq);
ajAlignSetScoreI(align, len);
/* ungapped alignment means same length
* for both sequences
*/
ajAlignSetSubRange(align, targetstart, 1, len,
ajSeqIsReversed(targetseq),
ajSeqGetLen(targetseq),
querystart, 1, len,
ajSeqIsReversed(queryseq),
ajSeqGetLen(queryseq));
}
}
if(dumpAlign)
{
ajAlignWrite(align);
ajAlignReset(align);
}
if(ajListGetLength(matchlist[i])>0 && dumpFeature)
{
embWordMatchListConvToFeat(matchlist[i],
&seqsetftables[i],
&seqallseqftable,
targetseq, queryseq);
ajFeattableWrite(ftoutforseqallseq, seqallseqftable);
ajFeattableDel(&seqallseqftable);
}
ajListIterDel(&iter);
}
embWordMatchListDelete(&matchlist[i]);
}
}
/* search completed, now report statistics */
for(i=0;i<npatterns;i++)
{
sumAllScore += wordsw[i]->lenMatches;
for(j=0;j<wordsw[i]->nseqs;j++)
nmatchesseqset[wordsw[i]->seqindxs[j]] +=
wordsw[i]->nSeqMatches[j];
}
ajFmtPrintF(logfile, "Number of sequences in the file scanned "
"for patterns: %u\n", ajSeqallGetCount(seqall));
ajFmtPrintF(logfile, "Number of all matches: %Lu"
" (wordmatch finds exact matches only)\n", nAllMatches);
if(nAllMatches>0)
{
ajFmtPrintF(logfile, "Sum of match lengths: %Lu\n", sumAllScore);
ajFmtPrintF(logfile, "Average match length: %.2f\n",
sumAllScore*1.0/nAllMatches);
ajFmtPrintF(logfile, "\nDistribution of the matches among pattern"
" sequences:\n");
ajFmtPrintF(logfile, "-----------------------------------------"
"-----------\n");
for(i=0;i<ajSeqsetGetSize(seqset);i++)
{
if (nmatchesseqset[i]>0)
ajFmtPrintF(logfile, "%-42s: %8u\n",
ajSeqGetNameC(ajSeqsetGetseqSeq(seqset, i)),
nmatchesseqset[i]);
ajFeattableWrite(ftoutforseqsetseq, seqsetftables[i]);
ajFeattableDel(&seqsetftables[i]);
}
ajFmtPrintF(logfile, "\nPattern statistics:\n");
ajFmtPrintF(logfile, "-------------------\n");
if(wordlen>7)
ajStrAppendCountK(&padding, ' ', wordlen-7);
paddedheader = ajFmtString(header,padding);
ajFmtPrintF(logfile, paddedheader);
for(i=0;i<npatterns;i++)
if (wordsw[i]->nMatches>0)
ajFmtPrintF(logfile, "%-7s: %12u %12u %17.2f\n",
wordsw[i]->word->fword, wordsw[i]->nseqs,
wordsw[i]->nMatches,
wordsw[i]->lenMatches*1.0/wordsw[i]->nMatches);
}
}
for(i=0;i<npatterns;i++)
{
for(j=0;j<wordsw[i]->nseqs;j++)
AJFREE(wordsw[i]->locs[j]);
AJFREE(wordsw[i]->locs);
AJFREE(wordsw[i]->seqindxs);
AJFREE(wordsw[i]->nnseqlocs);
AJFREE(wordsw[i]->nSeqMatches);
AJFREE(wordsw[i]);
}
embWordFreeTable(&wordsTable);
AJFREE(wordsw);
AJFREE(matchlist);
AJFREE(lastlocation);
AJFREE(nmatchesseqset);
AJFREE(seqsetftables);
if(dumpAlign)
{
ajAlignClose(align);
ajAlignDel(&align);
}
if(dumpFeature)
{
ajFeattabOutDel(&ftoutforseqsetseq);
ajFeattabOutDel(&ftoutforseqallseq);
}
ajFileClose(&logfile);
ajSeqallDel(&seqall);
ajSeqsetDel(&seqset);
ajSeqDel(&queryseq);
ajStrDel(&padding);
AJFREE(paddedheader);
embExit();
return 0;
}
