int main(int argc, char **argv)
{
AjPSeqall seqall;
AjPSeq seq = NULL;
AjPFile primerFile; /* read the primer pairs from a file */
AjPFile outf;
AjPList primerList;
ajint mmp = 0;
embInit("primersearch", argc, argv);
seqall = ajAcdGetSeqall("seqall");
outf = ajAcdGetOutfile("outfile");
primerFile = ajAcdGetInfile("infile");
mmp = ajAcdGetInt("mismatchpercent");
/* build list of forward/reverse primer pairs as read from primerfile */
primerList = ajListNew();
/* read in primers from primerfile, classify and compile them */
primersearch_read_primers(&primerList,primerFile, mmp);
/* check there are primers to be searched */
if(!ajListGetLength(primerList))
{
ajErr("No suitable primers found - exiting");
embExitBad();
return 0;
}
/* query sequences one by one */
while(ajSeqallNext(seqall,&seq))
primersearch_primer_search(primerList, seq);
/* output the results */
primersearch_print_hits(primerList, outf);
/* delete all nodes of list, then the list itself */
ajListMap(primerList, primersearch_free_primer, NULL);
ajListFree(&primerList);
ajListFree(&primerList);
ajFileClose(&outf);
ajSeqallDel(&seqall);
ajSeqDel(&seq);
ajFileClose(&primerFile);
embExit();
return 0;
}
static ajint supermatcher_findstartpoints(AjPTable seq1MatchTable,
const AjPSeq b,
const AjPSeq a, ajint *start1,
ajint *start2, ajint *end1,
ajint *end2)
{
ajint max = -10;
ajint offset = 0;
AjPList matchlist = NULL;
AjPList ordered = NULL;
ajint amax;
ajint bmax;
ajint bega;
ajint begb;
amax = ajSeqGetLen(a)-1;
bmax = ajSeqGetLen(b)-1;
bega = ajSeqGetOffset(a);
begb = ajSeqGetOffset(b);
ajDebug("supermatcher_findstartpoints len %d %d off %d %d\n",
amax, bmax, bega, begb);
matchlist = embWordBuildMatchTable(seq1MatchTable, b, ajTrue);
if(!matchlist)
return 0;
else if(!matchlist->Count)
{
embWordMatchListDelete(&matchlist);
return 0;
}
/* order and add if the gap is gapmax or less */
/* create list header bit*/
ordered = ajListNew();
supermatcher_orderandconcat(matchlist, ordered);
/* this sets global structure conmax to point to a matchlist element */
ajListMap(ordered,supermatcher_findmax, &max);
ajDebug("findstart conmax off:%d count:%d total:%d\n",
conmax->offset, conmax->count, conmax->total,
ajListGetLength(conmax->list));
offset = conmax->offset;
/* the offset is all we needed! we can delete everything */
ajListMap(ordered,supermatcher_removelists, NULL);
ajListFree(&ordered);
embWordMatchListDelete(&matchlist); /* free the match structures */
if(offset > 0)
{
*start1 = offset;
*start2 = 0;
}
else
{
*start2 = 0-offset;
*start1 = 0;
}
*end1 = *start1;
*end2 = *start2;
ajDebug("++ end1 %d -> %d end2 %d -> %d\n", *end1, amax, *end2, bmax);
while(*end1<amax && *end2<bmax)
{
(*end1)++;
(*end2)++;
}
ajDebug("++ end1 %d end2 %d\n", *end1, *end2);
ajDebug("supermatcher_findstartpoints has %d..%d [%d] %d..%d [%d]\n",
*start1, *end1, ajSeqGetLen(a), *start2, *end2, ajSeqGetLen(b));
return 1;
}
static void supermatcher_orderandconcat(AjPList list,AjPList ordered)
{
ajListMap(list,supermatcher_matchListOrder, ordered);
return;
}
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;
}
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)
{
AjPSeqall seqall;
AjPFile primfile;
AjPStr rdline = NULL;
Primer primdata;
AjPStrTok handle = NULL;
AjPList primList = NULL;
embInit("stssearch", argc, argv);
primfile = ajAcdGetInfile("infile");
out = ajAcdGetOutfile("outfile");
seqall = ajAcdGetSeqall("seqall");
while(ajReadlineTrim(primfile, &rdline))
{
if(ajStrGetCharFirst(rdline) == '#')
continue;
if(ajStrSuffixC(rdline, ".."))
continue;
AJNEW(primdata);
primdata->Name = NULL;
primdata->Oligoa = NULL;
primdata->Oligob = NULL;
handle = ajStrTokenNewC(rdline, " \t");
ajStrTokenNextParse(&handle, &primdata->Name);
if(!(nprimers % 1000))
ajDebug("Name [%d]: '%S'\n", nprimers, primdata->Name);
ajStrTokenNextParse(&handle, &primdata->Oligoa);
ajStrFmtUpper(&primdata->Oligoa);
primdata->Prima = ajRegComp(primdata->Oligoa);
ajStrTokenNextParse(&handle, &primdata->Oligob);
ajStrFmtUpper(&primdata->Oligob);
primdata->Primb = ajRegComp(primdata->Oligob);
ajStrTokenDel(&handle);
if(!nprimers)
primList = ajListNew();
ajListPushAppend(primList, primdata);
nprimers++;
}
if(!nprimers)
ajFatal("No primers read\n");
ajDebug("%d primers read\n", nprimers);
while(ajSeqallNext(seqall, &seq))
{
ajSeqFmtUpper(seq);
ajStrAssignS(&seqstr, ajSeqGetSeqS(seq));
ajStrAssignS(&revstr, ajSeqGetSeqS(seq));
ajSeqstrReverse(&revstr);
ajDebug("Testing: %s\n", ajSeqGetNameC(seq));
ntests = 0;
ajListMap(primList, stssearch_primTest, NULL);
}
ajFileClose(&out);
ajSeqallDel(&seqall);
ajSeqDel(&seq);
ajFileClose(&out);
ajStrDel(&revstr);
ajStrDel(&seqstr);
ajFileClose(&primfile);
ajListMap(primList, stssearch_primDel, NULL);
ajListFree(&primList);
ajStrDel(&rdline);
embExit();
return 0;
}
