int main(int argc, char **argv)
{
AjPSeqall seqall;
AjPSeq seq = NULL;
AjPReport report = NULL;
AjPStr jaspdir = NULL;
AjPStr menu = NULL;
AjPStr substr = NULL;
AjPStr mats = NULL;
AjPStr excl = NULL;
float thresh = 0.;
ajuint recurs = 0;
AjPStr dir = NULL;
AjPStr mfname = NULL;
AjPList flist = NULL;
AjPList hits = NULL;
AjPStr head = NULL;
ajint begin;
ajint end;
ajuint mno;
char cp;
ajuint i;
AjPTable mattab = NULL;
AjPFeattable TabRpt = NULL;
AjBool both = ajFalse;
embInit("jaspscan", argc, argv);
seqall = ajAcdGetSeqall("sequence");
menu = ajAcdGetListSingle("menu");
mats = ajAcdGetString("matrices");
excl = ajAcdGetString("exclude");
thresh = ajAcdGetFloat("threshold");
report = ajAcdGetReport("outfile");
both = ajAcdGetBoolean("both");
jaspdir = ajStrNew();
substr = ajStrNew();
flist = ajListNew();
hits = ajListNew();
dir = ajStrNew();
head = ajStrNew();
cp = ajStrGetCharFirst(menu);
if(cp=='C')
ajStrAssignC(&jaspdir,J_COR);
else if(cp=='F')
ajStrAssignC(&jaspdir,J_FAM);
else if(cp=='P')
ajStrAssignC(&jaspdir,J_PHY);
else if(cp=='N')
ajStrAssignC(&jaspdir,J_CNE);
else if(cp=='O')
ajStrAssignC(&jaspdir,J_POL);
else if(cp=='S')
ajStrAssignC(&jaspdir,J_SPL);
else
ajFatal("Invalid JASPAR database selection");
ajStrAssignS(&dir, ajDatafileValuePath());
if(!ajStrGetLen(dir))
ajFatal("EMBOSS DATA directory couldn't be determined");
jaspscan_ParseInput(dir, jaspdir, mats, excl, &recurs, flist);
mno = ajListGetLength(flist);
if(cp == 'C')
mattab = jaspscan_ReadCoreList(jaspdir);
if(cp == 'F')
mattab = jaspscan_ReadFamList(jaspdir);
if(cp == 'P')
mattab = jaspscan_ReadCoreList(jaspdir);
if(cp == 'N')
mattab = jaspscan_ReadCoreList(jaspdir);
if(cp == 'O')
mattab = jaspscan_ReadCoreList(jaspdir);
if(cp == 'S')
mattab = jaspscan_ReadCoreList(jaspdir);
ajFmtPrintS(&head,"Database scanned: %S Threshold: %.3f",jaspdir,thresh);
ajReportSetHeaderS(report,head);
while(ajSeqallNext(seqall, &seq))
{
begin = ajSeqallGetseqBegin(seqall);
end = ajSeqallGetseqEnd(seqall);
ajStrAssignSubC(&substr,ajSeqGetSeqC(seq),begin-1,end-1);
ajStrFmtUpper(&substr);
TabRpt = ajFeattableNewSeq(seq);
for(i=0; i < mno; ++i)
{
ajListPop(flist,(void **)&mfname);
jaspscan_scan(substr,begin,mfname, cp, thresh, both, hits);
ajListPushAppend(flist, (void **)mfname);
}
jaspscan_ReportHits(TabRpt,mattab,hits);
ajReportWrite(report, TabRpt, seq);
ajFeattableDel(&TabRpt);
}
while(ajListPop(flist,(void **)&mfname))
ajStrDel(&mfname);
ajStrDel(&dir);
ajStrDel(&menu);
ajStrDel(&excl);
ajStrDel(&substr);
ajStrDel(&mats);
ajStrDel(&head);
ajStrDel(&jaspdir);
ajSeqDel(&seq);
ajTableMapDel(mattab,jaspscan_ClearTable,NULL);
ajTableFree(&mattab);
ajListFree(&flist);
ajListFree(&hits);
ajSeqallDel(&seqall);
ajReportDel(&report);
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)
{
AjPSeq seq = NULL;
AjPReport report = NULL;
AjPFeattable feat=NULL;
AjPStr sstr = NULL;
const AjPStr sname = NULL;
AjPStr revcomp = NULL;
ajint RStotal;
AjPStr enzymes = NULL; /* string for RE selection */
AjPList relist = NULL;
ajint begin;
ajint end;
ajint radj;
ajint start;
AjBool sshow;
AjBool tshow;
AjBool allmut;
AjPList results1 = NULL; /* for forward strand */
AjPList results2 = NULL; /* for reverse strand */
AjPList shits;
AjPList nshits;
AjPStr tailstr = NULL;
embInit("silent", argc, argv);
seq = ajAcdGetSeq("sequence");
enzymes = ajAcdGetString("enzymes");
sshow = ajAcdGetBoolean("sshow");
tshow = ajAcdGetBoolean("tshow");
allmut = ajAcdGetBoolean("allmut");
report = ajAcdGetReport ("outfile");
shits = ajListNew();
nshits = ajListNew();
/*calling function to read in RE info*/
RStotal = silent_restr_read(&relist,enzymes);
begin = ajSeqGetBegin(seq); /* returns the seq start posn, or 1
if no start has been set */
end = ajSeqGetEnd(seq); /* returns the seq end posn, or seq
length if no end has been set */
radj=begin+end+1; /* posn adjustment for complementary
strand */
ajStrAssignSubC(&sstr,ajSeqGetSeqC(seq),--begin,--end);
ajStrFmtUpper(&sstr);
sname = ajSeqGetNameS(seq);
ajStrAssignC(&revcomp,ajStrGetPtr(sstr));
ajSeqstrReverse(&revcomp);
start = begin+1;
feat = ajFeattableNewDna(ajSeqGetNameS(seq));
if(sshow)
{
silent_fmt_sequence("SEQUENCE", sstr,&tailstr,start,ajTrue);
}
results1 = silent_mismatch(sstr,relist,&tailstr,sname,RStotal,begin,radj,
ajFalse,end,tshow);
silent_split_hits(&results1,&shits,&nshits,allmut);
ajReportSetHeaderC(report,
"KEY:\n"
"EnzymeName: Enzyme name\n"
"RS-Pattern: Restriction enzyme recognition site "
"pattern\n"
"Base-Posn: Position of base to be mutated\n"
"AAs: Amino acid. Original sequence(.)After mutation\n"
"Silent: Yes for unchanged amino acid\n"
"Mutation: The base mutation to perform\n\n"
"Creating silent and non-silent mutations\n");
silent_fmt_hits(shits,feat, ajTrue, ajFalse);
if(allmut)
{
silent_fmt_hits(nshits,feat, ajFalse, ajFalse);
}
if(sshow)
{
silent_fmt_sequence("REVERSE SEQUENCE", revcomp,&tailstr,start,ajTrue);
}
results2 = silent_mismatch(revcomp,relist,&tailstr,
sname,RStotal,begin,radj,
ajTrue,end,tshow);
silent_split_hits(&results2,&shits,&nshits,allmut);
silent_fmt_hits(shits,feat, ajTrue, ajTrue);
if(allmut)
{
silent_fmt_hits(nshits,feat, ajFalse, ajTrue);
}
ajReportSetStatistics(report, 1, ajSeqGetLenTrimmed(seq));
ajReportSetTailS(report, tailstr);
(void) ajReportWrite (report,feat,seq);
ajFeattableDel(&feat);
ajStrDel(&revcomp);
ajStrDel(&enzymes);
ajListFree(&results1);
ajListFree(&results2);
ajListFree(&shits);
ajListFree(&nshits);
ajReportClose(report);
ajReportDel(&report);
ajSeqDel(&seq);
ajStrDel(&sstr);
silent_relistdel(&relist);
ajStrDel(&tailstr);
embExit();
return 0;
}
int main(int argc, char **argv)
{
AjPSeqall seqall;
AjPSeq seq = NULL;
AjPReport outf = NULL;
AjPFile inf = NULL;
ajint begin;
ajint end;
AjPList l = NULL;
AjPStr strand = NULL;
AjPStr substr = NULL;
AjPStr line = NULL;
AjPStr name = NULL;
AjPStr acc = NULL;
AjPStr bf = NULL;
AjPStr menu;
AjPStr pattern = NULL;
AjPStr opattern = NULL;
AjPStr pname = NULL;
AjPStr key = NULL;
AjPStr value = NULL;
AjPTable atable = NULL;
AjPTable btable = NULL;
ajint mismatch;
ajint minlength;
ajint sum;
ajint v;
char cp;
const char *p;
embInit("tfscan", argc, argv);
seqall = ajAcdGetSeqall("sequence");
outf = ajAcdGetReport("outfile");
mismatch = ajAcdGetInt("mismatch");
minlength = ajAcdGetInt("minlength");
menu = ajAcdGetListSingle("menu");
pname = ajStrNew();
cp=ajStrGetCharFirst(menu);
if(cp=='F')
ajStrAssignC(&pname,"tffungi");
else if(cp=='I')
ajStrAssignC(&pname,"tfinsect");
else if(cp=='O')
ajStrAssignC(&pname,"tfother");
else if(cp=='P')
ajStrAssignC(&pname,"tfplant");
else if(cp=='V')
ajStrAssignC(&pname,"tfvertebrate");
else if(cp=='C')
inf = ajAcdGetDatafile("custom");
if(cp!='C')
{
inf = ajDatafileNewInNameS(pname);
if(!inf)
ajFatal("Either EMBOSS_DATA undefined or TFEXTRACT needs running");
}
name = ajStrNew();
acc = ajStrNew();
bf = ajStrNewC("");
substr = ajStrNew();
line = ajStrNew();
pattern = ajStrNewC("AA");
opattern = ajStrNew();
while(ajSeqallNext(seqall, &seq))
{
begin=ajSeqallGetseqBegin(seqall);
end=ajSeqallGetseqEnd(seqall);
ajStrAssignC(&name,ajSeqGetNameC(seq));
strand=ajSeqGetSeqCopyS(seq);
ajStrAssignSubC(&substr,ajStrGetPtr(strand),begin-1,end-1);
ajStrFmtUpper(&substr);
l=ajListNew();
atable = ajTablestrNew(1000);
btable = ajTablestrNew(1000);
sum=0;
while(ajReadlineTrim(inf,&line))
{
p = ajStrGetPtr(line);
if(!*p || *p=='#' || *p=='\n' || *p=='!')
continue;
ajFmtScanS(line,"%S%S%S",&pname,&pattern,&acc);
p += ajStrGetLen(pname);
while(*p && *p==' ')
++p;
p += ajStrGetLen(pattern);
while(*p && *p==' ')
++p;
p += ajStrGetLen(acc);
while(*p && *p==' ')
++p;
ajStrAssignS(&opattern,pattern);
ajStrAssignC(&bf,p); /* rest of line */
v = embPatVariablePattern(pattern,substr,pname,l,0,
mismatch,begin);
if(v)
{
key = ajStrNewS(pname);
value = ajStrNewS(acc);
ajTablePut(atable,(void *)key,(void *)value);
key = ajStrNewS(pname);
value = ajStrNewS(bf);
ajTablePut(btable,(void *)key,(void *)value);
}
sum += v;
}
if(sum)
tfscan_print_hits(&l,sum,outf,atable,seq,minlength,
btable);
ajFileSeek(inf,0L,0);
ajListFree(&l);
ajTablestrFree(&atable);
ajTablestrFree(&btable);
ajStrDel(&strand);
}
ajStrDel(&line);
ajStrDel(&name);
ajStrDel(&acc);
ajStrDel(&pname);
ajStrDel(&opattern);
ajStrDel(&bf);
ajStrDel(&pattern);
ajStrDel(&substr);
ajSeqDel(&seq);
ajFileClose(&inf);
ajReportClose(outf);
ajReportDel(&outf);
ajSeqallDel(&seqall);
ajSeqDel(&seq);
ajStrDel(&menu);
embExit();
return 0;
}
int main(int argc, char **argv)
{
AjPSeqall seqall;
AjPSeq a;
AjPStr substr;
AjPStr rname;
ajint be;
ajint en;
ajint len;
AjBool unfavoured;
AjBool overlap;
AjBool allpartials;
AjPStr menu;
AjPStr rag;
ajint n = 0;
ajint r = 0;
AjPFile outf = NULL;
AjPReport report = NULL;
AjPFeattable TabRpt = NULL;
AjPStr tmpStr = NULL;
AjPList l;
AjPList pa;
AjPFile mfptr = NULL;
AjBool nterm = ajFalse;
AjBool cterm = ajFalse;
AjBool dorag = ajFalse;
ajint ncomp;
ajint npart;
EmbPPropMolwt *mwdata = NULL;
AjBool mono;
embInit("digest", argc, argv);
seqall = ajAcdGetSeqall("seqall");
menu = ajAcdGetListSingle("menu");
dorag = ajAcdGetBoolean("ragging");
rag = ajAcdGetListSingle("termini");
unfavoured = ajAcdGetBoolean("unfavoured");
overlap = ajAcdGetBoolean("overlap");
allpartials = ajAcdGetBoolean("allpartials");
report = ajAcdGetReport("outfile");
mfptr = ajAcdGetDatafile("mwdata");
mono = ajAcdGetBoolean("mono");
/* obsolete. Can be uncommented in acd file and here to reuse */
/* outf = ajAcdGetOutfile("originalfile"); */
ajStrToInt(menu, &n);
--n;
ajStrToInt(rag, &r);
if(r==2 || r==4)
nterm = ajTrue;
if(r==3 || r==4)
cterm = ajTrue;
mwdata = embPropEmolwtRead(mfptr);
while(ajSeqallNext(seqall, &a))
{
substr = ajStrNew();
be = ajSeqGetBegin(a);
en = ajSeqGetEnd(a);
ajStrAssignSubC(&substr,ajSeqGetSeqC(a),be-1,en-1);
ajStrFmtUpper(&substr);
len = en-be+1;
l = ajListNew();
pa = ajListNew();
rname = ajStrNew();
TabRpt = ajFeattableNewSeq(a);
embPropCalcFragments(ajStrGetPtr(substr),n,&l,&pa,
unfavoured,overlap,
allpartials,&ncomp,&npart,&rname,
nterm, cterm, dorag, mwdata, mono);
if(outf)
ajFmtPrintF(outf,"DIGEST of %s from %d to %d Molwt=%10.3f\n\n",
ajSeqGetNameC(a),be,en,
embPropCalcMolwt(ajSeqGetSeqC(a),0,len-1,mwdata,mono));
if(!ncomp)
{
if(outf)
ajFmtPrintF(outf,
"Is not proteolytically digested using %s\n",
ajStrGetPtr(rname));
}
else
{
if(outf)
{
ajFmtPrintF(outf,"Complete digestion with %s "
"yields %d fragments:\n",
ajStrGetPtr(rname),ncomp);
digest_print_hits(l,outf,be,ajStrGetPtr(substr));
}
ajFmtPrintS(&tmpStr,
"Complete digestion with %S yields %d fragments",
rname,ncomp);
ajReportSetHeaderS(report, tmpStr);
digest_report_hits(TabRpt,l,be, ajStrGetPtr(substr));
ajReportWrite(report, TabRpt, a);
ajFeattableClear(TabRpt);
}
if(overlap && !allpartials && npart)
{
if(outf)
{
ajFmtPrintF(outf,
"\n\nPartial digest with %s yields %d extras.\n",
ajStrGetPtr(rname),npart);
ajFmtPrintF(outf,"Only overlapping partials shown:\n");
digest_print_hits(pa,outf,be,ajStrGetPtr(substr));
}
ajFmtPrintS(&tmpStr,
"\n\nPartial digest with %S yields %d extras.\n",
rname,npart);
ajFmtPrintAppS(&tmpStr,"Only overlapping partials shown:\n");
ajReportSetHeaderS(report, tmpStr);
digest_report_hits(TabRpt, pa,be,ajStrGetPtr(substr));
ajReportWrite(report, TabRpt, a);
ajFeattableClear(TabRpt);
}
if(allpartials && npart)
{
if(outf)
{
ajFmtPrintF(outf,
"\n\nPartial digest with %s yields %d extras.\n",
ajStrGetPtr(rname),npart);
ajFmtPrintF(outf,"All partials shown:\n");
digest_print_hits(pa,outf,be,ajStrGetPtr(substr));
}
ajFmtPrintS(&tmpStr,
"\n\nPartial digest with %S yields %d extras.\n",
rname,npart);
ajFmtPrintAppS(&tmpStr,"All partials shown:\n");
ajReportSetHeaderS(report, tmpStr);
digest_report_hits(TabRpt, pa,be, ajStrGetPtr(substr));
ajReportWrite(report, TabRpt, a);
ajFeattableClear(TabRpt);
}
}
embPropMolwtDel(&mwdata);
ajReportDel(&report);
ajFeattableDel(&TabRpt);
ajSeqDel(&a);
ajSeqallDel(&seqall);
ajStrDel(&rname);
ajStrDel(&substr);
ajListFree(&pa);
ajListFree(&l);
ajStrDel(&menu);
ajStrDel(&rag);
if(outf)
ajFileClose(&outf);
ajFileClose(&mfptr);
ajStrDel(&tmpStr);
embExit();
return 0;
}
