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;
}
int main(int argc, char **argv)
{
AjPFile inf = NULL;
AjPFile inf2 = NULL;
AjPFeattable tab = NULL;
AjPReport report = NULL;
AjPSeq sequence = NULL;
AjPStr redatanew = NULL;
AjPStr str = NULL;
AjPStr regexp = NULL;
AjPStr temp = NULL;
AjPStr text = NULL;
AjPStr docdata = NULL;
AjPStr data = NULL;
AjPStr accession = NULL;
AjPStr name = NULL;
EmbPPatMatch match = NULL;
AjPStr savereg = NULL;
AjPStr fthit = NULL;
AjBool full;
AjBool prune;
ajint i;
ajint number;
ajint start;
ajint end;
ajint length;
ajint zstart;
ajint zend;
const char *p;
ajint seqlength;
AjPStr tmpstr = NULL;
AjPStr tailstr = NULL;
AjPFeature gf;
embInit("patmatmotifs", argc, argv);
ajStrAssignC(&fthit, "SO:0001067");
savereg = ajStrNew();
str = ajStrNew();
regexp = ajStrNew();
temp = ajStrNew();
data = ajStrNew();
accession = ajStrNew();
text = ajStrNew();
name = ajStrNew();
sequence = ajAcdGetSeq("sequence");
report = ajAcdGetReport("outfile");
full = ajAcdGetBoolean("full");
prune = ajAcdGetBoolean("prune");
ajSeqFmtUpper(sequence); /* prosite regexs are all upper case */
tab = ajFeattableNewSeq(sequence);
ajStrAssignC(&tailstr, "");
seqlength = ajStrGetLen(str);
str = ajSeqGetSeqCopyS(sequence);
redatanew = ajStrNewC("PROSITE/prosite.lines");
docdata = ajStrNewC("PROSITE/");
inf = ajDatafileNewInNameS(redatanew);
if(!inf)
ajFatal("Either EMBOSS_DATA undefined or PROSEXTRACT needs running");
ajFmtPrintAppS(&tmpstr, "Full: %B\n", full);
ajFmtPrintAppS(&tmpstr, "Prune: %B\n", prune);
ajFmtPrintAppS(&tmpstr, "Data_file: %F\n", inf);
ajReportSetHeaderS(report, tmpstr);
while(ajReadlineTrim(inf, ®exp))
{
p=ajStrGetPtr(regexp);
if(*p && *p!=' ' && *p!='^')
{
p=ajSysFuncStrtok(p," ");
ajStrAssignC(&name,p);
if(prune)
if(ajStrMatchCaseC(name,"myristyl") ||
ajStrMatchCaseC(name,"asn_glycosylation") ||
ajStrMatchCaseC(name,"camp_phospho_site") ||
ajStrMatchCaseC(name,"pkc_phospho_site") ||
ajStrMatchCaseC(name,"ck2_phospho_site") ||
ajStrMatchCaseC(name,"tyr_phospho_site"))
{
for(i=0;i<4;++i)
ajReadlineTrim(inf, ®exp);
continue;
}
p=ajSysFuncStrtok(NULL," ");
ajStrAssignC(&accession,p);
}
if(ajStrPrefixC(regexp, "^"))
{
p = ajStrGetPtr(regexp);
ajStrAssignC(&temp,p+1);
ajStrAssignC(&savereg,p+1);
match = embPatMatchFind(temp, str, ajFalse, ajFalse);
number = embPatMatchGetNumber(match);
for(i=0; i<number; i++)
{
seqlength = ajStrGetLen(str);
start = 1+embPatMatchGetStart(match, i);
end = 1+embPatMatchGetEnd(match, i);
length = embPatMatchGetLen(match, i);
gf = ajFeatNew(tab, NULL, fthit, start, end,
(float) length, ' ', 0);
ajFmtPrintS(&tmpstr, "*motif %S", name);
ajFeatTagAddSS(gf, NULL, tmpstr);
if(start-5<0)
zstart = 0;
else
zstart = start-5;
if(end+5> seqlength)
zend = end;
else
zend = end+5;
ajStrAssignSubS(&temp, str, zstart, zend);
}
if(full && number)
{
ajStrAssignC(&redatanew,ajStrGetPtr(docdata));
ajStrAppendC(&redatanew,ajStrGetPtr(accession));
inf2 = ajDatafileNewInNameS(redatanew);
if(!inf2)
continue;
/*
** Insert Prosite documentation from files made by
** prosextract.c
*/
ajFmtPrintAppS(&tailstr, "Motif: %S\n", name);
ajFmtPrintAppS(&tailstr, "Count: %d\n\n", number);
while(ajReadlineTrim(inf2, &text))
ajFmtPrintAppS(&tailstr, "%S\n", text);
ajFmtPrintAppS(&tailstr, "\n***************\n\n");
ajFileClose(&inf2);
}
embPatMatchDel(&match);
}
}
ajReportSetTailS(report,tailstr);
ajReportWrite(report, tab, sequence);
ajReportDel(&report);
ajFeattableDel(&tab);
ajStrDel(&temp);
ajStrDel(®exp);
ajStrDel(&savereg);
ajStrDel(&str);
ajStrDel(&data);
ajStrDel(&docdata);
ajStrDel(&text);
ajStrDel(&redatanew);
ajStrDel(&accession);
ajSeqDel(&sequence);
ajStrDel(&tailstr);
ajStrDel(&fthit);
ajStrDel(&name);
ajStrDel(&tmpstr);
ajFeattableDel(&tab);
ajFileClose(&inf);
embExit();
return 0;
}
int main(int argc, char **argv)
{
AjPSeqall seqall;
AjPSeq seq = NULL;
AjPStr seqcmp = NULL;
AjPStr enzymes = NULL;
AjPFile outf = NULL;
ajint begin;
ajint end;
ajint min;
ajint max;
ajint sitelen;
AjBool alpha;
AjBool single;
AjBool blunt;
AjBool ambiguity;
AjBool sticky;
AjBool plasmid;
AjBool threeprime;
AjBool commercial;
AjBool html;
AjBool limit;
AjBool frags;
AjBool methyl;
AjPFile dfile;
AjPFile enzfile = NULL;
AjPFile equfile = NULL;
AjPFile methfile = NULL;
AjPStr name = NULL;
AjPTable table = NULL;
ajint hits;
AjPList l = NULL;
embInit("restover", argc, argv);
seqall = ajAcdGetSeqall("sequence");
seqcmp = ajAcdGetString("seqcomp");
ajStrFmtUpper(&seqcmp);
outf = ajAcdGetOutfile("outfile");
/*
** Some of these are not needed but I left them in case someone wants to
** use them some time ...
*/
enzymes = ajStrNewC("all");
min = ajAcdGetInt("min");
max = ajAcdGetInt("max");
sitelen = 2;
threeprime = ajAcdGetBoolean("threeprime");
blunt = ajAcdGetBoolean("blunt");
sticky = ajAcdGetBoolean("sticky");
single = ajAcdGetBoolean("single");
html = ajAcdGetBoolean("html");
alpha = ajAcdGetBoolean("alphabetic");
ambiguity = ajAcdGetBoolean("ambiguity");
plasmid = ajAcdGetBoolean("plasmid");
commercial = ajAcdGetBoolean("commercial");
limit = ajAcdGetBoolean("limit");
frags = ajAcdGetBoolean("fragments");
methyl = ajAcdGetBoolean("methylation");
dfile = ajAcdGetDatafile("datafile");
methfile = ajAcdGetDatafile("mfile");
if(single)
max = min = 1;
table = ajTablestrNew(EQUGUESS);
l = ajListNew();
if(threeprime)
ajStrReverse(&seqcmp);
/* read the local file of enzymes names */
restover_read_file_of_enzyme_names(&enzymes);
if(!dfile)
{
enzfile = ajDatafileNewInNameC(ENZDATA);
if(!enzfile)
ajFatal("Cannot locate enzyme file. Run REBASEEXTRACT");
}
else
{
enzfile = dfile;
}
if(limit)
{
equfile = ajDatafileNewInNameC(EQUDATA);
if(!equfile)
limit=ajFalse;
else
{
restover_read_equiv(equfile,table);
ajFileClose(&equfile);
}
}
while(ajSeqallNext(seqall, &seq))
{
begin = ajSeqallGetseqBegin(seqall);
end = ajSeqallGetseqEnd(seqall);
ajFileSeek(enzfile,0L,0);
ajSeqFmtUpper(seq);
hits = embPatRestrictMatch(seq,begin,end,enzfile,methfile,enzymes,
sitelen,plasmid,ambiguity,min,max,blunt,
sticky,commercial,methyl,l);
ajDebug("hits:%d listlen:%u\n", hits, ajListGetLength(l));
if(hits)
{
name = ajStrNewC(ajSeqGetNameC(seq));
restover_printHits(seq, seqcmp, outf,l,name,hits,begin,end,
min,max,plasmid,
sitelen,limit,table,alpha,frags,
html);
ajStrDel(&name);
}
ajListFree(&l);
}
ajListFree(&l);
ajSeqDel(&seq);
ajFileClose(&outf);
ajFileClose(&dfile);
ajFileClose(&enzfile);
ajFileClose(&equfile);
ajFileClose(&methfile);
ajSeqallDel(&seqall);
ajStrDel(&seqcmp);
ajStrDel(&enzymes);
ajStrDel(&name);
ajTablestrFree(&table);
embExit();
return 0;
}
