/* @funcstatic domainalign_writeid ********************************************
**
** House-keeping function.
**
** @param [r] domain [AjPDomain] Domain.
** @param [r] noden [ajint] Node number.
** @param [r] daf [AjPDirout] Domain alignment files.
** @param [r] super [AjPDirout] Structural superimposition files.
** @param [r] align [AjPStr *] Align.
** @param [r] alignc [AjPStr *] Alignc.
**
** @return [void]
**
** @@
****************************************************************************/
static void domainalign_writeid(AjPDomain domain,
ajint noden,
AjPDirout daf,
AjPDirout super,
AjPStr *align,
AjPStr *alignc)
{
AjPStr temp=NULL;
temp=ajStrNew();
if(noden==1)
ajStrFromInt(&temp, domain->Scop->Sunid_Class);
else if (noden==2)
ajStrFromInt(&temp, domain->Scop->Sunid_Fold);
else if (noden==3)
ajStrFromInt(&temp, domain->Scop->Sunid_Superfamily);
else if (noden==4)
ajStrFromInt(&temp, domain->Scop->Sunid_Family);
else if (noden==5)
ajStrFromInt(&temp, domain->Cath->Class_Id );
else if (noden==6)
ajStrFromInt(&temp, domain->Cath->Arch_Id);
else if (noden==7)
ajStrFromInt(&temp, domain->Cath->Topology_Id);
else if (noden==8)
ajStrFromInt(&temp, domain->Cath->Superfamily_Id);
else if (noden==9)
ajStrFromInt(&temp, domain->Cath->Family_Id);
ajStrAssignS(align, temp);
ajStrInsertS(align, 0, ajDiroutGetPath(daf));
ajStrAppendC(align, ".");
ajStrAppendS(align, ajDiroutGetExt(daf));
ajStrAssignS(alignc, temp);
ajStrInsertS(alignc, 0, ajDiroutGetPath(super));
ajStrAppendC(alignc, ".");
ajStrAppendS(alignc, ajDiroutGetExt(super));
ajDebug("daf file '%S' super file '%S'\n", *align, *alignc);
ajStrDel(&temp);
return;
}
AjPResquery ajResqueryNewResquery(const AjPResquery qry)
{
AjPResquery ret;
AJNEW0(ret);
if(!qry)
return ret;
ajStrAssignS(&ret->Datatype, qry->Datatype);
ajStrAssignS(&ret->Format, qry->Format);
ajStrAssignS(&ret->FormatTerm, qry->FormatTerm);
ajStrAssignS(&ret->Term, qry->Term);
ajStrAssignS(&ret->Url, qry->Url);
ret->Nterms = qry->Nterms;
return ret;
}
/* @funcstatic domainalign_writelast ******************************************
**
** House-keeping function.
**
** @param [r] domain [AjPDomain] Domain.
** @param [r] noden [ajint] Node number.
** @param [r] last_node [AjPStr *] Last node.
** @param [r] last_nodeid [ajint *] Id of last node.
**
** @return [void]
**
** @@
****************************************************************************/
static void domainalign_writelast(AjPDomain domain,
ajint noden,
AjPStr *last_node,
ajint *last_nodeid)
{
if(noden==1)
{
ajStrAssignS(last_node, domain->Scop->Class);
*last_nodeid = domain->Scop->Sunid_Class;
}
else if (noden==2)
{
ajStrAssignS(last_node, domain->Scop->Fold);
*last_nodeid = domain->Scop->Sunid_Fold;
}
else if (noden==3)
{
ajStrAssignS(last_node, domain->Scop->Superfamily);
*last_nodeid = domain->Scop->Sunid_Superfamily;
}
else if (noden==4)
{
ajStrAssignS(last_node, domain->Scop->Family);
*last_nodeid = domain->Scop->Sunid_Family;
}
else if (noden==5)
{
ajStrAssignS(last_node, domain->Cath->Class);
*last_nodeid = domain->Cath->Class_Id ;
}
else if (noden==6)
{
ajStrAssignS(last_node, domain->Cath->Architecture);
*last_nodeid = domain->Cath->Arch_Id;
}
else if (noden==7)
{
ajStrAssignS(last_node, domain->Cath->Topology);
*last_nodeid = domain->Cath->Topology_Id;
}
else if (noden==8)
{
ajStrAssignS(last_node, domain->Cath->Superfamily);
*last_nodeid = domain->Cath->Superfamily_Id;
}
else if (noden==9)
{
/* There is no text describing the CATH families. */
ajFmtPrintS(last_node, "%d", domain->Cath->Family_Id);
*last_nodeid = domain->Cath->Family_Id;
}
}
static void jaspextract_readmatrixlist(AjPTable mtable, const AjPStr directory)
{
const AjPStr datadir = NULL;
AjPStr matrixfile = NULL;
AjPFile inf = NULL;
AjPStr line = NULL;
AjPStr key = NULL;
AjPStr value = NULL;
matrixfile = ajStrNew();
datadir = ajDatafileValuePath();
if(!datadir)
ajFatal("jaspextract: Cannot determine the EMBOSS data directory");
ajFmtPrintS(&matrixfile,"%S%s",directory,MATRIXFILE);
if(!ajFilenameExistsRead(matrixfile))
ajFatal("jaspextract: Directory (%S) doesn't appear to be a JASPAR "
"one\nNo matrix_list.txt file found",directory);
inf = ajFileNewInNameS(matrixfile);
if(!inf)
ajFatal("Cannot open input file: %S",matrixfile);
while(ajReadline(inf,&line))
{
key = ajStrNew();
if(ajFmtScanS(line,"%S",&key) != 1)
{
ajStrDel(&key);
continue;
}
value = ajStrNew();
ajStrAssignS(&value,line);
ajTablePut(mtable,(void *)key, (void *)value);
}
ajFileClose(&inf);
ajStrDel(&matrixfile);
ajStrDel(&line);
return;
}
static void dbiblast_newname(AjPStr* nname, const AjPStr oname,
const char *suff)
{
ajStrAssignS(nname, oname);
if(ajStrGetCharFirst(oname)=='@')
ajStrCutStart(nname, 1);
ajStrAppendK(nname, '.');
ajStrAppendC(nname, suff);
return;
}
static void notseq_readfile(const AjPStr exclude, AjPStr *pattern)
{
AjPFile file = NULL;
AjPStr line;
AjPStr filename = NULL;
const char *p = NULL;
if(ajStrFindC(exclude, "@") != 0)
{
ajStrAssignS(pattern, exclude);
}
else
{
ajStrAssignS(&filename, exclude);
ajStrTrimC(&filename, "@"); /* remove the @ */
file = ajFileNewInNameS(filename);
if(file == NULL)
ajFatal("Cannot open the file of sequence names: '%S'", filename);
/* blank off the file name and replace with the sequence names */
ajStrSetClear(pattern);
line = ajStrNew();
while(ajReadlineTrim(file, &line))
{
p = ajStrGetPtr(line);
if(!*p || *p == '#' || *p == '!')
continue;
ajStrAppendS(pattern, line);
ajStrAppendC(pattern, ",");
}
ajStrDel(&line);
ajStrDel(&filename);
ajFileClose(&file);
}
return;
}
static void eprimer3_write_primer(AjPFile outfile, const char *tag,
const AjPStr pos,
const AjPStr tm, const AjPStr gc,
const AjPStr seq,
AjBool rev, ajint begin)
{
ajint startint;
ajint lenint;
float tmfloat;
float gcfloat;
AjPStr start = NULL;
AjPStr lenstr = NULL;
ajlong comma;
if(ajStrGetLen(pos))
{
ajStrToFloat(tm, &tmfloat);
ajStrToFloat(gc, &gcfloat);
comma = ajStrFindC(pos, ",");
ajStrAssignS(&start, pos);
ajStrCutRange(&start, comma, ajStrGetLen(start)-1);
ajStrToInt(start, &startint);
startint += begin;
ajStrAssignS(&lenstr, pos);
ajStrCutRange(&lenstr, 0, comma);
ajStrToInt(lenstr, &lenint);
if(rev)
startint = startint - lenint + 1;
ajFmtPrintF(outfile, " %s %6d %4d %2.2f %2.2f %S\n\n",
tag, startint, lenint, tmfloat, gcfloat, seq);
}
ajStrDel(&start);
ajStrDel(&lenstr);
return;
}
void ajHistSetmultiXlabelS(AjPHist thys, ajuint indexnum, const AjPStr str)
{
if(indexnum >= thys->numofdatapoints)
{
ajErr("Histograms can only be allocated from 0 to %d. NOT %u",
thys->numofdatapoints-1,indexnum);
return;
}
ajStrAssignS(&thys->hists[indexnum]->xaxis, str);
return;
}
static void splitter_MakeSubSeqName (AjPStr * name_ptr,
const AjPSeq seq, ajuint start,
ajuint end)
{
AjPStr value = ajStrNew();
/* create a nice name for the subsequence */
ajStrAssignS(name_ptr, ajSeqGetNameS(seq));
ajStrAppendC(name_ptr, "_");
ajStrFromUint(&value, ajSeqGetBegin(seq)+start);
ajStrAppendS(name_ptr, value);
ajStrAppendC(name_ptr, "-");
ajStrFromUint(&value, ajSeqGetBegin(seq)+end);
ajStrAppendS(name_ptr, value);
ajStrDel(&value);
}
static void emira_dobools(AjPStr *cl, AjPTable table)
{
ajuint i;
AjBool bqual = ajFalse;
AjBool def = ajFalse;
AjPStr prefix = NULL;
AjPStr key = NULL;
AjPStr value = NULL;
prefix = ajStrNew();
key = ajStrNew();
i = 0;
while(mirabools[i].qname)
{
bqual = ajAcdGetBoolean(mirabools[i].qname);
ajStrAssignC(&key,mirabools[i].qname);
ajStrAssignC(&prefix,"");
value = ajTableFetch(table, key);
if(value)
ajStrAssignS(&prefix,value);
def = ajCharMatchC(mirabools[i].def,"N") ? ajFalse : ajTrue;
if(bqual != def)
{
ajFmtPrintAppS(cl," -%S%s=",prefix,mirabools[i].mname);
if(bqual)
ajFmtPrintAppS(cl,"yes");
else
ajFmtPrintAppS(cl,"no");
}
++i;
}
ajStrDel(&key);
ajStrDel(&prefix);
return;
}
static void emira_doinfiles(AjPStr *cl, AjPTable table)
{
ajuint i;
AjPStr squal = NULL;
AjPStr prefix = NULL;
AjPStr key = NULL;
AjPStr value = NULL;
AjPFile infile = NULL;
prefix = ajStrNew();
key = ajStrNew();
i = 0;
while(mirainfiles[i].qname)
{
infile = ajAcdGetInfile(mirainfiles[i].qname);
if(infile)
squal = ajStrNewS(ajFileGetName(infile));
else
squal = ajStrNewS(ajAcdGetValueDefault(mirainfiles[i].qname));
ajStrAssignC(&key,mirainfiles[i].mname);
ajStrAssignC(&prefix,"");
value = ajTableFetch(table, key);
if(value)
ajStrAssignS(&prefix,value);
if(!ajStrMatchC(squal,mirainfiles[i].def))
ajFmtPrintAppS(cl," -%S%s=%S",prefix,mirainfiles[i].mname,
squal);
ajStrDel(&squal);
ajFileClose(&infile);
++i;
}
ajStrDel(&key);
ajStrDel(&prefix);
return;
}
static void filebuffLineAdd(AjPFilebuff buff, const AjPStr line)
{
/* ajDebug("fileBuffLineAdd '%S'\n", line);*/
if(buff->Freelines)
{
if(!buff->Lines)
{
/* Need to set first line in list */
buff->Lines = buff->Freelines;
}
else
buff->Last->Next = buff->Freelines;
buff->Last = buff->Freelines;
buff->Freelines = buff->Freelines->Next;
if(!buff->Freelines)
{
/* Free list now empty */
buff->Freelast = NULL;
}
}
else
{
/* No Free list, make a new string */
if(!buff->Lines)
buff->Lines = AJNEW0(buff->Last);
else
buff->Last = AJNEW0(buff->Last->Next);
}
ajStrAssignS(&buff->Last->Line, line);
buff->Prev = buff->Curr;
buff->Curr = buff->Last;
buff->Last->Next = NULL;
buff->Last->Fpos = buff->Fpos;
buff->Pos++;
buff->Size++;
return;
}
static PMemFile dbiblast_memfopenfile(const AjPStr name)
{
PMemFile ret;
AjPFile fp;
fp = ajFileNewInNameS(name);
if(!fp)
return NULL;
AJNEW0(ret);
ajStrAssignS(&ret->Name, name);
ret->IsMem = 0;
ret->File = fp;
ret->Size = 0;
ret->Mem = NULL;
ajDebug("fopened '%S'\n", name);
return ret;
}
static void emira_dointegers(AjPStr *cl, AjPTable table)
{
ajuint i;
ajint iqual;
int dval = 0;
AjPStr prefix = NULL;
AjPStr key = NULL;
AjPStr value = NULL;
prefix = ajStrNew();
key = ajStrNew();
i = 0;
while(miraintegers[i].qname)
{
iqual = ajAcdGetInt(miraintegers[i].qname);
ajStrAssignC(&key,miraintegers[i].qname);
ajStrAssignC(&prefix,"");
value = ajTableFetch(table, key);
if(value)
ajStrAssignS(&prefix,value);
if(sscanf(miraintegers[i].def,"%d",&dval) != 1)
ajFatal("Dval conversion error in emira_dointegers()");
if(iqual != dval)
ajFmtPrintAppS(cl," -%S%s=%d",prefix,miraintegers[i].mname,iqual);
++i;
}
ajStrDel(&key);
ajStrDel(&prefix);
return;
}
static void dbiblast_dbname(AjPStr* dbname,
const AjPStr oname, const char *suff)
{
AjPStr suffix = NULL;
ajFmtPrintS(&suffix, ".%s", suff);
ajStrAssignS(dbname, oname);
if(ajStrGetCharFirst(oname)=='@')
ajStrCutStart(dbname, 1);
if(!ajStrSuffixS(*dbname, suffix))
{
ajStrDel(&suffix);
return;
}
ajStrCutEnd(dbname, ajStrGetLen(suffix));
ajStrDel(&suffix);
return;
}
static void jaspscan_ParseInput(const AjPStr dir, const AjPStr jaspdir,
const AjPStr mats, const AjPStr excl,
ajuint *recurs, AjPList ret)
{
ajuint nm = 0;
ajuint ne = 0;
AjPStr *carr = NULL;
AjPStr *earr = NULL;
AjPFile inf = NULL;
AjPStr line = NULL;
AjPStr comm = NULL;
AjPStr val = NULL;
ajuint i;
ajuint j;
char c;
ajuint rlen = 0;
if(*recurs > JASPSCAN_RECURS)
ajFatal("Too many recursion levels in matrix list files");
line = ajStrNew();
comm = ajStrNew();
if(mats)
{
nm = ajArrCommaList(mats,&carr);
for(i=0; i < nm; ++i)
{
if(ajStrGetCharFirst(carr[i]) != '@')
ajStrFmtUpper(&carr[i]);
if(ajStrMatchC(carr[i],"ALL"))
{
jaspscan_GetFileList(dir, jaspdir, "*", ret);
ajListSortUnique(ret, ajStrVcmp, jaspscan_strdel);
}
else if(ajStrGetCharFirst(carr[i]) == '@')
{
ajStrTrimStartC(&carr[i],"@");
inf = ajFileNewInNameS(carr[i]);
if(!inf)
ajFatal("Cannot open list file %S",carr[i]);
while(ajReadlineTrim(inf,&line))
{
ajStrRemoveWhite(&line);
c = ajStrGetCharFirst(line);
if(c == '#' || c== '!')
continue;
if(ajStrGetLen(comm))
ajStrAppendC(&comm,",");
ajStrFmtUpper(&line);
ajStrAppendS(&comm,line);
}
*recurs += 1;
jaspscan_ParseInput(dir,jaspdir,comm,NULL,recurs,ret);
*recurs -= 1;
ajListSortUnique(ret, ajStrVcmp, jaspscan_strdel);
ajFileClose(&inf);
}
else
{
jaspscan_GetFileList(dir,jaspdir,ajStrGetPtr(carr[i]),ret);
ajListSortUnique(ret, ajStrVcmp, jaspscan_strdel);
}
}
for(i=0; i < nm; ++i)
ajStrDel(&carr[i]);
AJFREE(carr);
}
if(excl)
{
ne = ajArrCommaList(excl,&earr);
for(i=0; i < ne; ++i)
{
if(ajStrGetCharFirst(earr[i]) != '@')
ajStrFmtUpper(&earr[i]);
if(ajStrGetCharFirst(earr[i]) == '@')
{
ajStrTrimStartC(&earr[i],"@");
inf = ajFileNewInNameS(earr[i]);
if(!inf)
ajFatal("Cannot open list file %S",earr[i]);
while(ajReadlineTrim(inf,&line))
{
ajStrRemoveWhite(&line);
c = ajStrGetCharFirst(line);
if(c == '#' || c== '!')
continue;
if(ajStrGetLen(comm))
ajStrAppendC(&comm,",");
ajStrFmtUpper(&line);
ajStrAppendS(&comm,line);
}
*recurs += 1;
jaspscan_ParseInput(dir,jaspdir,NULL,comm,recurs,ret);
*recurs -= 1;
ajListSortUnique(ret, ajStrVcmp, jaspscan_strdel);
ajFileClose(&inf);
}
else
{
ajStrAssignS(&line,earr[i]);
ajStrAppendC(&line,J_EXT);
rlen = ajListGetLength(ret);
for(j=0; j < rlen; ++j)
{
ajListPop(ret,(void **)&val);
if(ajStrSuffixS(val,line))
ajStrDel(&val);
else
ajListPushAppend(ret,(void *)val);
}
}
}
for(i=0; i < ne; ++i)
ajStrDel(&earr[i]);
AJFREE(earr);
}
ajStrDel(&line);
ajStrDel(&comm);
return;
}
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;
}
static void jaspscan_coretoken(PJspmat info, const AjPStr str)
{
const char *p;
const char *q;
AjPStr key = NULL;
AjPStr value = NULL;
value = ajStrNew();
key = ajStrNew();
p = ajStrGetPtr(str);
q = p;
while(*q != '"' && *q != '\t')
++q;
--q;
ajStrAssignSubC(&key,p,0,q-p-1);
while(*q != '"')
++q;
if(*(q+1) != '"')
{
p = q + 1;
++q;
while(*q != '"')
++q;
ajStrAssignSubC(&value,p,0,q-p-1);
if(ajStrMatchC(key,"acc"))
ajStrAssignS(&info->protseq,value);
if(ajStrMatchC(key,"medline"))
ajStrAssignS(&info->pmid,value);
if(ajStrMatchC(key,"species"))
ajStrAssignS(&info->species,value);
if(ajStrMatchC(key,"sysgroup"))
ajStrAssignS(&info->sgroup,value);
if(ajStrMatchC(key,"total_ic"))
ajStrAssignS(&info->content,value);
if(ajStrMatchC(key,"type"))
ajStrAssignS(&info->exp,value);
if(ajStrMatchC(key,"included_models"))
ajStrAssignS(&info->models,value);
if(ajStrMatchC(key,"MCS"))
ajStrAssignS(&info->mcs,value);
if(ajStrMatchC(key,"jaspar"))
ajStrAssignS(&info->jaspar,value);
if(ajStrMatchC(key,"transfac"))
ajStrAssignS(&info->transfac,value);
if(ajStrMatchC(key,"Description"))
ajStrAssignS(&info->desc,value);
if(ajStrMatchC(key,"description"))
ajStrAssignS(&info->desc,value);
if(ajStrMatchC(key,"comment"))
ajStrAssignS(&info->comment,value);
if(ajStrMatchC(key,"End relative to TSS"))
ajStrAssignS(&info->erttss,value);
if(ajStrMatchC(key,"Start relative to TSS"))
ajStrAssignS(&info->srttss,value);
if(ajStrMatchC(key,"consensus"))
ajStrAssignS(&info->consens,value);
}
ajStrDel(&key);
ajStrDel(&value);
return;
}
int main(int argc, char **argv)
{
ajint begin, end;
AjPSeqall seqall;
AjPSeq seq;
EmbPShow ss;
AjPFile outfile;
AjPStr tablename;
ajint table;
AjPRange uppercase;
AjPRange highlight;
AjBool threeletter;
AjBool numberseq;
AjBool nameseq;
ajint width;
ajint length;
ajint margin;
AjBool description;
ajint offset;
AjBool html;
AjPStr descriptionline;
ajint orfminsize;
AjPTrn trnTable;
AjBool translation;
AjBool reverse;
AjBool cutlist;
AjBool flat;
EmbPMatMatch mm = NULL;
AjPStr *framelist;
AjBool frames[6]; /* frames to be translated 1 to 3, -1 to -3 */
/* stuff for tables and lists of enzymes and hits */
ajint default_mincuts = 1;
ajint default_maxcuts = 2000000000;
AjPTable hittable; /* enzyme hits */
/* stuff lifted from Alan's 'restrict.c' */
AjPStr enzymes = NULL;
ajint mincuts;
ajint maxcuts;
ajint sitelen;
AjBool single;
AjBool blunt;
AjBool sticky;
AjBool ambiguity;
AjBool plasmid;
AjBool commercial;
AjBool limit;
AjBool methyl;
AjPFile enzfile = NULL;
AjPFile equfile = NULL;
AjPFile methfile = NULL;
AjPTable retable = NULL;
ajint hits;
AjPList restrictlist = NULL;
embInit("remap", argc, argv);
seqall = ajAcdGetSeqall("sequence");
outfile = ajAcdGetOutfile("outfile");
tablename = ajAcdGetListSingle("table");
uppercase = ajAcdGetRange("uppercase");
highlight = ajAcdGetRange("highlight");
threeletter = ajAcdGetBoolean("threeletter");
numberseq = ajAcdGetBoolean("number");
width = ajAcdGetInt("width");
length = ajAcdGetInt("length");
margin = ajAcdGetInt("margin");
nameseq = ajAcdGetBoolean("name");
description = ajAcdGetBoolean("description");
offset = ajAcdGetInt("offset");
html = ajAcdGetBoolean("html");
orfminsize = ajAcdGetInt("orfminsize");
translation = ajAcdGetBoolean("translation");
reverse = ajAcdGetBoolean("reverse");
cutlist = ajAcdGetBoolean("cutlist");
flat = ajAcdGetBoolean("flatreformat");
framelist = ajAcdGetList("frame");
/* restriction enzyme stuff */
mincuts = ajAcdGetInt("mincuts");
maxcuts = ajAcdGetInt("maxcuts");
sitelen = ajAcdGetInt("sitelen");
single = ajAcdGetBoolean("single");
blunt = ajAcdGetBoolean("blunt");
sticky = ajAcdGetBoolean("sticky");
ambiguity = ajAcdGetBoolean("ambiguity");
plasmid = ajAcdGetBoolean("plasmid");
commercial = ajAcdGetBoolean("commercial");
limit = ajAcdGetBoolean("limit");
enzymes = ajAcdGetString("enzymes");
methfile = ajAcdGetDatafile("mfile");
methyl = ajAcdGetBoolean("methylation");
if(!blunt && !sticky)
ajFatal("Blunt/Sticky end cutters shouldn't both be disabled.");
/* get the number of the genetic code used */
ajStrToInt(tablename, &table);
trnTable = ajTrnNewI(table);
/* read the local file of enzymes names */
remap_read_file_of_enzyme_names(&enzymes);
/* get the frames to be translated */
remap_GetFrames(framelist, frames);
while(ajSeqallNext(seqall, &seq))
{
/* get begin and end positions */
begin = ajSeqGetBegin(seq)-1;
end = ajSeqGetEnd(seq)-1;
/* do the name and description */
if(nameseq)
{
if(html)
ajFmtPrintF(outfile, "<H2>%S</H2>\n",
ajSeqGetNameS(seq));
else
ajFmtPrintF(outfile, "%S\n", ajSeqGetNameS(seq));
}
if(description)
{
/*
** wrap the description line at the width of the sequence
** plus margin
*/
if(html)
ajFmtPrintF(outfile, "<H3>%S</H3>\n",
ajSeqGetDescS(seq));
else
{
descriptionline = ajStrNew();
ajStrAssignS(&descriptionline, ajSeqGetDescS(seq));
ajStrFmtWrap(&descriptionline, width+margin);
ajFmtPrintF(outfile, "%S\n", descriptionline);
ajStrDel(&descriptionline);
}
}
/* get the restriction cut sites */
/*
** most of this is lifted from the program 'restrict.c' by Alan
** Bleasby
*/
if(single)
maxcuts=mincuts=1;
retable = ajTablestrNew(EQUGUESS);
enzfile = ajDatafileNewInNameC(ENZDATA);
if(!enzfile)
ajFatal("Cannot locate enzyme file. Run REBASEEXTRACT");
if(limit)
{
equfile = ajDatafileNewInNameC(EQUDATA);
if(!equfile)
limit = ajFalse;
else
remap_read_equiv(&equfile, &retable, commercial);
}
ajFileSeek(enzfile, 0L, 0);
restrictlist = ajListNew();
/* search for hits, but don't use mincuts and maxcuts criteria yet */
hits = embPatRestrictMatch(seq, begin+1, end+1, enzfile, methfile,
enzymes, sitelen,plasmid, ambiguity,
default_mincuts, default_maxcuts, blunt,
sticky, commercial, methyl,
restrictlist);
ajDebug("Remap found %d hits\n", hits);
if(hits)
{
/* this bit is lifted from printHits */
embPatRestrictRestrict(restrictlist, hits, !limit,
ajFalse);
if(limit)
remap_RestrictPreferred(restrictlist,retable);
}
ajFileClose(&enzfile);
ajFileClose(&methfile);
/*
** Remove those violating the mincuts and maxcuts
** criteria, but save them in hittable for printing out later.
** Keep a count of how many hits each enzyme gets in hittable.
*/
hittable = ajTablestrNewCase(TABLEGUESS);
remap_RemoveMinMax(restrictlist, hittable, mincuts, maxcuts);
/* make the Show Object */
ss = embShowNew(seq, begin, end, width, length, margin, html, offset);
if(html)
ajFmtPrintF(outfile, "<PRE>");
/* create the format to display */
embShowAddBlank(ss);
embShowAddRE(ss, 1, restrictlist, plasmid, flat);
embShowAddSeq(ss, numberseq, threeletter, uppercase, highlight);
if(!numberseq)
embShowAddTicknum(ss);
embShowAddTicks(ss);
if(reverse)
{
embShowAddComp(ss, numberseq);
embShowAddRE(ss, -1, restrictlist, plasmid, flat);
}
if(translation)
{
if(reverse)
embShowAddBlank(ss);
if(frames[0])
embShowAddTran(ss, trnTable, 1, threeletter,
numberseq, NULL, orfminsize,
AJFALSE, AJFALSE, AJFALSE, AJFALSE);
if(frames[1])
embShowAddTran(ss, trnTable, 2, threeletter,
numberseq, NULL, orfminsize,
AJFALSE, AJFALSE, AJFALSE, AJFALSE);
if(frames[2])
embShowAddTran(ss, trnTable, 3, threeletter,
numberseq, NULL, orfminsize,
AJFALSE, AJFALSE, AJFALSE, AJFALSE);
if(reverse)
{
embShowAddTicks(ss);
if(frames[5])
embShowAddTran(ss, trnTable, -3, threeletter,
numberseq, NULL, orfminsize,
AJFALSE, AJFALSE, AJFALSE, AJFALSE);
if(frames[4])
embShowAddTran(ss, trnTable, -2, threeletter,
numberseq, NULL, orfminsize,
AJFALSE, AJFALSE, AJFALSE, AJFALSE);
if(frames[3])
embShowAddTran(ss, trnTable, -1, threeletter,
numberseq, NULL, orfminsize,
AJFALSE, AJFALSE, AJFALSE, AJFALSE);
}
}
embShowPrint(outfile, ss);
/* display a list of the Enzymes that cut and don't cut */
if(cutlist)
{
remap_CutList(outfile, hittable,
limit, html, mincuts, maxcuts);
remap_NoCutList(outfile, hittable, html, enzymes, blunt,
sticky, sitelen, commercial, ambiguity,
limit, retable);
}
/* add a gratuitous newline at the end of the sequence */
ajFmtPrintF(outfile, "\n");
/* tidy up */
embShowDel(&ss);
while(ajListPop(restrictlist,(void **)&mm))
embMatMatchDel(&mm);
ajListFree(&restrictlist);
remap_DelTable(&hittable);
ajTablestrFree(&retable);
}
ajTrnDel(&trnTable);
ajSeqallDel(&seqall);
ajSeqDel(&seq);
ajFileClose(&outfile);
ajStrDel(&tablename);
ajStrDel(&enzymes);
ajStrDelarray(&framelist);
ajRangeDel(&uppercase);
ajRangeDel(&highlight);
embExit();
return 0;
}
static void remap_NoCutList(AjPFile outfile, const AjPTable hittable,
AjBool html, const AjPStr enzymes, AjBool blunt,
AjBool sticky, ajuint sitelen, AjBool commercial,
AjBool ambiguity, AjBool limit,
const AjPTable retable)
{
/* for iterating over hittable */
PValue value;
void **keyarray = NULL; /* array for table */
void **valarray = NULL; /* array for table */
ajint i;
/* list of enzymes that cut */
AjPList cutlist;
AjIList citer; /* iterator for cutlist */
AjPStr cutname = NULL;
AjBool found;
/* for parsing value->iso string */
AjPStrTok tok;
char tokens[] = " ,";
AjPStr code = NULL;
const char *p;
/* for reading in enzymes names */
AjPFile enzfile = NULL;
AjPStr *ea;
ajint ne; /* number of enzymes */
AjBool isall = ajTrue;
/* list of enzymes that don't cut */
AjPList nocutlist;
AjIList niter; /* iterator for nocutlist */
AjPStr nocutname = NULL;
/* count of rejected enzymes not matching criteria */
ajint rejected_count = 0;
EmbPPatRestrict enz;
/* for renaming preferred isoschizomers */
AjPList newlist;
/*
**
** Make a list of enzymes('cutlist') that hit
** including the isoschizomer names
**
*/
ajDebug("Make a list of all enzymes that cut\n");
cutlist = ajListstrNew();
nocutlist = ajListstrNew();
ajTableToarrayKeysValues(hittable, &keyarray, &valarray);
for(i = 0; keyarray[i]; i++)
{
value = (PValue) valarray[i];
cutname = ajStrNew();
ajStrAssignRef(&cutname, keyarray[i]);
ajListstrPushAppend(cutlist, cutname);
/* Add to cutlist all isoschizomers of enzymes that cut */
ajDebug("Add to cutlist all isoschizomers of enzymes that cut\n");
/* start token to parse isoschizomers names */
tok = ajStrTokenNewC(value->iso, tokens);
while(ajStrTokenNextParseC(&tok, tokens, &code))
{
cutname = ajStrNew();
ajStrAssignS(&cutname, code);
ajListstrPushAppend(cutlist, cutname);
}
ajStrTokenDel(&tok);
}
ajStrDel(&code);
AJFREE(keyarray);
AJFREE(valarray);
/*
** Read in list of enzymes ('nocutlist') - either all or
** the input enzyme list.
** Exclude those that don't match the selection criteria - count these.
*/
ajDebug("Read in a list of all input enzyme names\n");
ne = 0;
if(!enzymes)
isall = ajTrue;
else
{
/* get input list of enzymes into ea[] */
ne = ajArrCommaList(enzymes, &ea);
if(ajStrMatchCaseC(ea[0], "all"))
isall = ajTrue;
else
{
isall = ajFalse;
for(i=0; i<ne; ++i)
ajStrRemoveWhite(&ea[i]);
}
}
enzfile = ajDatafileNewInNameC(ENZDATA);
/* push all enzyme names without the required criteria onto nocutlist */
enz = embPatRestrictNew();
while(!ajFileIsEof(enzfile))
{
if(!embPatRestrictReadEntry(enz, enzfile))
continue;
/*
** If user entered explicit enzyme list, then check to see if
** this is one of that explicit list
*/
if(!isall)
{
found = AJFALSE;
for(i=0; i<ne; ++i)
if(ajStrMatchCaseS(ea[i], enz->cod))
{
found = AJTRUE;
break;
}
if(!found) /* not in the explicit list */
continue;
ajDebug("RE %S is in the input explicit list of REs\n", enz->cod);
}
/* ignore ncuts==0 as they are unknown */
if(!enz->ncuts)
{
/* number of cut positions */
ajDebug("RE %S has an unknown number of cut positions\n",
enz->cod);
continue;
}
ajDebug("RE %S has a known number of cut sites\n", enz->cod);
if(enz->len < sitelen)
{
/* recognition site length */
ajDebug("RE %S does not have a long enough recognition site\n",
enz->cod);
rejected_count++;
continue;
}
if(!blunt && enz->blunt)
{
/* blunt/sticky */
ajDebug("RE %S is blunt\n", enz->cod);
rejected_count++;
continue;
}
if(!sticky && !enz->blunt)
{
/* blunt/sticky */
ajDebug("RE %S is sticky\n", enz->cod);
rejected_count++;
continue;
}
/* commercially available enzymes have uppercase patterns */
p = ajStrGetPtr(enz->pat);
/*
** The -commercial qualifier is only used if we are searching
** through 'all' of the REBASE database - if we have specified an
** explicit list of enzymes then they are searched for whether or
** not they are commercially available
*/
if((*p >= 'a' && *p <= 'z') && commercial && isall)
{
ajDebug("RE %S is not commercial\n", enz->cod);
rejected_count++;
continue;
}
if(!ambiguity && remap_Ambiguous(enz->pat)) {
ajDebug("RE %S is ambiguous\n", enz->cod);
rejected_count++;
continue;
}
ajDebug("RE %S matches all required criteria\n", enz->cod);
code = ajStrNew();
ajStrAssignS(&code, enz->cod);
ajListstrPushAppend(nocutlist, code);
}
embPatRestrictDel(&enz);
ajFileClose(&enzfile);
for(i=0; i<ne; ++i)
if(ea[i])
ajStrDel(&ea[i]);
if(ne)
AJFREE(ea);
/*
** Change names of enzymes in the non-cutter list
** to that of preferred (prototype)
** enzyme name so that the isoschizomers of cutters
** will be removed from the
** non-cutter list in the next bit.
** Remove duplicate prototype names.
*/
if(limit)
{
newlist = ajListstrNew();
remap_RenamePreferred(nocutlist, retable, newlist);
ajListstrFreeData(&nocutlist);
nocutlist = newlist;
ajListSortUnique(nocutlist, remap_cmpcase, remap_strdel);
}
/*
** Iterate through the list of input enzymes removing those that are in
** the cutlist.
*/
ajDebug("Remove from the nocutlist all enzymes and isoschizomers "
"that cut\n");
/*
** This steps down both lists at the same time, comparing names and
** iterating to the next name in whichever list whose name compares
** alphabetically before the other. Where a match is found, the
** nocutlist item is deleted.
*/
ajListSort(nocutlist, remap_cmpcase);
ajListSort(cutlist, remap_cmpcase);
citer = ajListIterNewread(cutlist);
niter = ajListIterNew(nocutlist);
/*
while((cutname = (AjPStr)ajListIterGet(citer)) != NULL)
ajDebug("dbg cutname = %S\n", cutname);
*/
nocutname = (AjPStr)ajListIterGet(niter);
cutname = (AjPStr)ajListIterGet(citer);
ajDebug("initial cutname, nocutname: '%S' '%S'\n", cutname, nocutname);
while(nocutname != NULL && cutname != NULL)
{
i = ajStrCmpCaseS(cutname, nocutname);
ajDebug("compare cutname, nocutname: %S %S ", cutname, nocutname);
ajDebug("ajStrCmpCase=%d\n", i);
if(i == 0)
{
/* match - so remove from nocutlist */
ajDebug("ajListstrRemove %S\n", nocutname);
ajListstrIterRemove(niter);
nocutname = (AjPStr)ajListIterGet(niter);
/*
** Don't increment the cutname list pointer here
** - there may be more than one entry in the nocutname
** list with the same name because we have converted
** isoschizomers to their preferred name
*/
/* cutname = (AjPStr)ajListIterGet(citer); */
}
else if(i == -1)
/* cutlist name sorts before nocutlist name */
cutname = (AjPStr)ajListIterGet(citer);
else if(i == 1)
/* nocutlist name sorts before cutlist name */
nocutname = (AjPStr)ajListIterGet(niter);
}
ajListIterDel(&citer);
ajListIterDel(&niter);
ajListstrFreeData(&cutlist);
/* Print the resulting list of those that do not cut*/
ajDebug("Print out the list\n");
/* print the title */
if(html)
ajFmtPrintF(outfile, "<H2>");
ajFmtPrintF(outfile, "\n\n# Enzymes that do not cut\n\n");
if(html)
ajFmtPrintF(outfile, "</H2>\n");
if(html)
ajFmtPrintF(outfile, "<PRE>");
/* ajListSort(nocutlist, ajStrVcmp);*/
niter = ajListIterNewread(nocutlist);
i = 0;
while((nocutname = (AjPStr)ajListIterGet(niter)) != NULL)
{
ajFmtPrintF(outfile, "%-10S", nocutname);
/* new line after every 7 names printed */
if(i++ == 7)
{
ajFmtPrintF(outfile, "\n");
i = 0;
}
}
ajListIterDel(&niter);
/* end the output */
ajFmtPrintF(outfile, "\n");
if(html) {ajFmtPrintF(outfile, "</PRE>\n");}
/*
** Print the count of rejected enzymes
** N.B. This is the count of ALL rejected enzymes including all
** isoschizomers
*/
if(html)
ajFmtPrintF(outfile, "<H2>");
ajFmtPrintF(outfile,
"\n\n# No. of cutting enzymes which do not match the\n"
"# SITELEN, BLUNT, STICKY, COMMERCIAL, AMBIGUOUS criteria\n\n");
if(html)
ajFmtPrintF(outfile, "</H2>\n");
ajFmtPrintF(outfile, "%d\n", rejected_count);
ajDebug("Tidy up\n");
ajListstrFreeData(&nocutlist);
ajListstrFreeData(&cutlist);
return;
}
static void remap_RestrictPreferred(const AjPList l, const AjPTable t)
{
AjIList iter = NULL;
EmbPMatMatch m = NULL;
const AjPStr value = NULL;
AjPStr newiso = NULL;
AjBool found; /* name found in isoschizomer list */
/* for parsing value->iso string */
AjPStrTok tok = NULL;
char tokens[] = " ,";
AjPStr code = NULL;
iter = ajListIterNewread(l);
while((m = (EmbPMatMatch)ajListIterGet(iter)))
{
found = ajFalse;
/* get prototype name */
value = ajTableFetchS(t, m->cod);
if(value)
{
ajStrAssignC(&newiso, "");
/* parse isoschizomer names from m->iso */
ajStrTokenDel(&tok);
tok = ajStrTokenNewC(m->iso, tokens);
while(ajStrTokenNextParseC(&tok, tokens, &code))
{
if(ajStrGetLen(newiso) > 0)
ajStrAppendC(&newiso, ",");
/* found the prototype name? */
if(!ajStrCmpCaseS(code, value))
{
ajStrAppendS(&newiso, m->cod);
found = ajTrue;
}
else
ajStrAppendS(&newiso, code);
}
ajStrTokenDel(&tok);
/* if the name was not replaced, then add it in now */
if(!found)
{
if(ajStrGetLen(newiso) > 0)
ajStrAppendC(&newiso, ",");
ajStrAppendS(&newiso, m->cod);
}
ajDebug("RE: %S -> %S iso=%S newiso=%S\n", m->cod, value,
m->iso, newiso);
/* replace the old iso string with the new one */
ajStrAssignS(&m->iso, newiso);
/* rename the enzyme to the prototype name */
ajStrAssignS(&m->cod, value);
}
}
ajListIterDel(&iter);
ajStrDel(&newiso);
ajStrDel(&code);
ajStrTokenDel(&tok);
return;
}
int main(int argc, char *argv[])
{
embInitPV("gseqinfo", argc, argv, "GEMBASSY", "1.0.1");
struct soap soap;
AjPSeqall seqall;
AjPSeq seq;
AjPStr inseq = NULL;
AjPStr seqid = NULL;
AjPStr tmp = NULL;
AjPStr parse = NULL;
AjPStr numA = NULL;
AjPStr numT = NULL;
AjPStr numG = NULL;
AjPStr numC = NULL;
AjPStrTok handle = NULL;
ajint n;
char *in0;
char *result;
AjBool show = 0;
AjPFile outf = NULL;
seqall = ajAcdGetSeqall("sequence");
outf = ajAcdGetOutfile("outfile");
while(ajSeqallNext(seqall, &seq))
{
soap_init(&soap);
inseq = NULL;
ajStrAppendC(&inseq, ">");
ajStrAppendS(&inseq, ajSeqGetNameS(seq));
ajStrAppendC(&inseq, "\n");
ajStrAppendS(&inseq, ajSeqGetSeqS(seq));
ajStrAssignS(&seqid, ajSeqGetAccS(seq));
in0 = ajCharNewS(inseq);
if(soap_call_ns1__seqinfo(
&soap,
NULL,
NULL,
in0,
&result
) == SOAP_OK)
{
tmp = ajStrNewC(result);
ajStrExchangeCC(&tmp, "<", "\n");
ajStrExchangeCC(&tmp, ">", "\n");
handle = ajStrTokenNewC(tmp, "\n");
while(ajStrTokenNextParse(handle, &parse))
{
if(ajStrIsInt(parse))
if(!numA)
numA = ajStrNewS(parse);
else if(!numT)
numT = ajStrNewS(parse);
else if(!numG)
numG = ajStrNewS(parse);
else if(!numC)
numC = ajStrNewS(parse);
}
if(show)
ajFmtPrint("Sequence: %S A: %S T: %S G: %S C: %S\n",
seqid, numA, numT, numG, numC);
else
ajFmtPrintF(outf, "Sequence: %S A: %S T: %S G: %S C: %S\n",
seqid, numA, numT, numG, numC);
}
else
{
soap_print_fault(&soap, stderr);
}
soap_destroy(&soap);
soap_end(&soap);
soap_done(&soap);
AJFREE(in0);
ajStrDel(&inseq);
}
ajSeqallDel(&seqall);
ajSeqDel(&seq);
ajStrDel(&seqid);
embExit();
return 0;
}
int main(int argc, char *argv[])
{
embInitPV("gp2", argc, argv, "GEMBASSY", "1.0.1");
AjPSeqall seqall;
AjPSeq seq;
AjPStr inseq = NULL;
AjBool accid = ajFalse;
AjPStr restid = NULL;
AjPStr seqid = NULL;
AjPStr base = NULL;
AjPStr url = NULL;
AjPFile tmpfile = NULL;
AjPStr tmpname = NULL;
AjPFile outf = NULL;
seqall = ajAcdGetSeqall("sequence");
accid = ajAcdGetBoolean("accid");
outf = ajAcdGetOutfile("outfile");
base = ajStrNewC("rest.g-language.org");
gAssignUniqueName(&tmpname);
while(ajSeqallNext(seqall, &seq))
{
inseq = NULL;
if(!accid)
{
if(gFormatGenbank(seq, &inseq))
{
tmpfile = ajFileNewOutNameS(tmpname);
if(!tmpfile)
{
ajDie("Output file (%S) open error\n", tmpname);
}
ajFmtPrintF(tmpfile, "%S", inseq);
ajFileClose(&tmpfile);
ajFmtPrintS(&url, "http://%S/upload/upl.pl", base);
gFilePostSS(url, tmpname, &restid);
ajStrDel(&url);
ajSysFileUnlinkS(tmpname);
}
else
{
ajDie("Sequence does not have features\n"
"Proceeding with sequence accession ID\n");
accid = ajTrue;
}
}
if(accid)
{
ajStrAssignS(&restid, ajSeqGetAccS(seq));
if(!ajStrGetLen(restid))
{
ajStrAssignS(&restid, ajSeqGetNameS(seq));
}
if(!ajStrGetLen(restid))
{
ajDie("No valid header information\n");
}
}
ajStrAssignS(&seqid, ajSeqGetAccS(seq));
url = ajStrNew();
ajFmtPrintS(&url, "http://%S/%S/P2/output=f/tag=gene", base, restid);
ajFmtPrintF(outf, "Sequence: %S\n", seqid);
if(!gFileOutURLS(url, &outf))
{
ajDie("Failed to download result from:\n%S\n", url);
}
ajStrDel(&url);
ajStrDel(&inseq);
}
ajFileClose(&outf);
ajSeqallDel(&seqall);
ajSeqDel(&seq);
ajStrDel(&seqid);
embExit();
return 0;
}
/* @funcstatic sigscanlig_score_ligands_site *********************************
**
** Writes score data for ligands (Lighit objects) from individual hits to
** signatures (Hit objects). Site score method is used. List of hit objects
** must be sorted by ligand type and site number.
**
** @param [r] hits [const AjPList] List of hit objects.
** @return [AjBool] True on success
** @@
******************************************************************************/
AjPList sigscanlig_score_ligands_site(AjPList hits)
{
AjPList ret = NULL;
AjIList iter = NULL; /* Iterator. */
EmbPHit hit = NULL;
AjPStr prev_ligand = NULL;
SigPLighit lighit = NULL;
ajint prevsn = -1; /* Previous site number */
float score = 0.0; /* Score for current ligand */
ajint nhits = 0; /* No. of hits (patches) for current
ligand. */
ajint nsites = 0; /* No. of sites for current ligand */
float score_site = 0.0; /* Score for this site. */
ajint patch_cnt = 0; /* No. of patches in current site. */
ret = ajListNew();
prev_ligand = ajStrNew();
iter = ajListIterNew(hits);
/* Hits are already sorted by ligid & site number */
while((hit = (EmbPHit) ajListIterGet(iter)))
{
/* ajFmtPrint("Current hit: %S (ligid: %S, sn: %d) score: %f\n",
hit->Acc, hit->Sig->Ligid, hit->Sig->sn, hit->Score); */
/* New site */
if(hit->Sig->sn != prevsn)
{
if(patch_cnt)
score_site /= patch_cnt;
score += score_site;
patch_cnt = 0;
score_site = 0;
}
/* New ligand */
if((!ajStrMatchS(hit->Sig->Ligid, prev_ligand)))
{
if(nsites)
score /= nsites;
if(lighit)
{
/* lighit->ns = snarr_siz; */
lighit->ns = nsites;
lighit->np = nhits;
lighit->score = score;
ajStrAssignS(&lighit->ligid, prev_ligand);
ajListPushAppend(ret, lighit);
}
lighit = sigscanlig_LighitNew();
nsites = 0;
nhits=0;
prevsn = -1;
score = 0;
}
/* Increment count of sites and hits/patches (for current
ligand) and patches (for current site) */
if(hit->Sig->sn != prevsn)
nsites++;
score_site += hit->Score;
nhits++;
patch_cnt++;
ajStrAssignS(&prev_ligand, hit->Sig->Ligid);
prevsn = hit->Sig->sn;
}
/* Process last hit */
if(patch_cnt)
score_site /= patch_cnt;
score += score_site;
if(nsites)
score /= nsites;
if(lighit)
{
/* lighit->ns = snarr_siz; */
lighit->ns = nsites;
lighit->np = nhits;
lighit->score = score;
ajStrAssignS(&lighit->ligid, prev_ligand);
ajListPushAppend(ret, lighit);
}
ajListSort(ret, sigscanlig_MatchinvScore);
ajListIterDel(&iter);
ajStrDel(&prev_ligand);
return ret;
}
AjPList sigscanlig_score_ligands_patch(AjPList hits)
{
AjPList ret = NULL;
AjIList iter = NULL; /* Iterator. */
EmbPHit hit = NULL;
AjPStr prev_ligand = NULL;
SigPLighit lighit = NULL;
float score = 0.0;
ajint nhits = 0; /* No. of hits (patches) for current ligand. */
ajint nsites = 0; /* No. of sites for current ligand. */
ajint prevsn = -1; /* Previous site number */
ret = ajListNew();
prev_ligand = ajStrNew();
iter = ajListIterNew(hits);
while((hit = (EmbPHit) ajListIterGet(iter)))
{
/* New ligand */
if((!ajStrMatchS(hit->Sig->Ligid, prev_ligand)))
{
if(nhits)
score /= nhits;
if(lighit)
{
lighit->ns = nsites;
lighit->np = nhits;
lighit->score = score;
ajStrAssignS(&lighit->ligid, prev_ligand);
ajListPushAppend(ret, lighit);
}
lighit = sigscanlig_LighitNew();
nsites = 0;
nhits=0;
prevsn = -1;
score = 0;
}
/* Increment count of sites and hits/patches (for current ligand)
and patches (for current site) */
if(hit->Sig->sn != prevsn)
nsites++;
score+= hit->Score;
nhits++;
ajStrAssignS(&prev_ligand, hit->Sig->Ligid);
prevsn = hit->Sig->sn;
}
/* Process last hit */
if(nhits)
score /= nhits;
if(lighit)
{
lighit->ns = nsites;
lighit->np = nhits;
lighit->score = score;
ajStrAssignS(&lighit->ligid, prev_ligand);
ajListPushAppend(ret, lighit);
}
ajListSort(ret, sigscanlig_MatchinvScore);
ajListIterDel(&iter);
ajStrDel(&prev_ligand);
return ret;
}
int main(int argc, char *argv[])
{
embInitPV("gbaseinformationcontent", argc, argv, "GEMBASSY", "1.0.1");
AjPSeqall seqall;
AjPSeq seq;
AjPStr inseq = NULL;
AjPStr position = 0;
ajint PatLen = 0;
ajint upstream = 0;
ajint downstream = 0;
AjBool accid = ajFalse;
AjPStr restid = NULL;
AjPStr seqid = NULL;
AjPStr base = NULL;
AjPStr url = NULL;
AjPFile tmpfile = NULL;
AjPStr tmpname = NULL;
AjBool plot = 0;
AjPFile outf = NULL;
AjPFilebuff buff = NULL;
AjPGraph mult = NULL;
gPlotParams gpp;
AjPStr title = NULL;
seqall = ajAcdGetSeqall("sequence");
position = ajAcdGetSelectSingle("position");
PatLen = ajAcdGetInt("patlen");
upstream = ajAcdGetInt("upstream");
downstream = ajAcdGetInt("downstream");
accid = ajAcdGetBoolean("accid");
plot = ajAcdGetToggle("plot");
outf = ajAcdGetOutfile("outfile");
mult = ajAcdGetGraphxy("graph");
base = ajStrNewC("rest.g-language.org");
gAssignUniqueName(&tmpname);
while(ajSeqallNext(seqall, &seq))
{
inseq = NULL;
if(!accid)
{
if(gFormatGenbank(seq, &inseq))
{
gAssignUniqueName(&tmpname);
tmpfile = ajFileNewOutNameS(tmpname);
if(!tmpfile)
{
ajFmtError("Output file (%S) open error\n", tmpname);
embExitBad();
}
ajFmtPrintF(tmpfile, "%S", inseq);
ajFileClose(&tmpfile);
ajFmtPrintS(&url, "http://%S/upload/upl.pl", base);
gFilePostSS(url, tmpname, &restid);
ajStrDel(&url);
ajSysFileUnlinkS(tmpname);
}
else
{
ajFmtError("Sequence does not have features\n"
"Proceeding with sequence accession ID\n");
accid = ajTrue;
}
}
if(accid)
{
ajStrAssignS(&seqid, ajSeqGetAccS(seq));
if(!ajStrGetLen(seqid))
{
ajStrAssignS(&seqid, ajSeqGetNameS(seq));
}
if(!ajStrGetLen(seqid))
{
ajFmtError("No valid header information\n");
embExitBad();
}
ajStrAssignS(&restid, seqid);
}
ajStrAssignS(&seqid, ajSeqGetAccS(seq));
url = ajStrNew();
ajFmtPrintS(&url, "http://%S/%S/base_information_content/position=%S/"
"PatLen=%d/upstream=%d/downstream=%d/output=f/tag=gene",
base, restid, position, PatLen, upstream, downstream);
if(plot)
{
title = ajStrNew();
ajStrAppendC(&title, argv[0]);
ajStrAppendC(&title, " of ");
ajStrAppendS(&title, seqid);
gpp.title = ajStrNewS(title);
gpp.xlab = ajStrNewC("position");
gpp.ylab = ajStrNewC("information content");
if(!gFilebuffURLS(url, &buff))
{
ajDie("File downloading error from:\n%S\n", url);
}
if(!gPlotFilebuff(buff, mult, &gpp))
{
ajDie("Error in plotting\n");
}
AJFREE(gpp.title);
AJFREE(gpp.xlab);
AJFREE(gpp.ylab);
ajStrDel(&title);
ajFilebuffDel(&buff);
}
else
{
ajFmtPrintF(outf, "Sequence: %S\n", seqid);
if(!gFileOutURLS(url, &outf))
{
ajDie("File downloading error from:\n%S\n", url);
}
}
}
ajFileClose(&outf);
ajSeqallDel(&seqall);
ajSeqDel(&seq);
ajStrDel(&seqid);
ajStrDel(&position);
embExit();
return 0;
}
int main(int argc, char *argv[])
{
embInitPV("ggcskew", argc, argv, "GEMBASSY", "1.0.3");
AjPSeqall seqall;
AjPSeq seq;
AjPStr inseq = NULL;
AjBool accid = ajFalse;
AjPStr restid = NULL;
AjPStr seqid = NULL;
AjPStr base = NULL;
AjPStr url = NULL;
AjPStr tmpname = NULL;
AjPSeqout tmpout = NULL;
ajint window = 0;
ajint slide = 0;
AjBool cumulative = 0;
AjBool at = 0;
AjBool purine = 0;
AjBool keto = 0;
AjBool plot = 0;
AjPFile outf = NULL;
AjPFilebuff buff = NULL;
AjPGraph mult = NULL;
gPlotParams gpp;
AjPStr title = NULL;
seqall = ajAcdGetSeqall("sequence");
window = ajAcdGetInt("window");
slide = ajAcdGetInt("slide");
cumulative = ajAcdGetBoolean("cumulative");
at = ajAcdGetBoolean("at");
purine = ajAcdGetBoolean("purine");
keto = ajAcdGetBoolean("keto");
plot = ajAcdGetToggle("plot");
outf = ajAcdGetOutfile("outfile");
mult = ajAcdGetGraphxy("graph");
base = ajStrNewC("rest.g-language.org");
gAssignUniqueName(&tmpname);
ajStrAppendC(&tmpname, ".fasta");
while(ajSeqallNext(seqall, &seq))
{
tmpout = ajSeqoutNew();
if(!ajSeqoutOpenFilename(tmpout, tmpname))
{
embExitBad();
}
ajSeqoutSetFormatS(tmpout,ajStrNewC("fasta"));
ajSeqoutWriteSeq(tmpout, seq);
ajSeqoutClose(tmpout);
ajSeqoutDel(&tmpout);
ajFmtPrintS(&url, "http://%S/upload/upl.pl", base);
gFilePostSS(url, tmpname, &restid);
ajStrDel(&url);
ajSysFileUnlinkS(tmpname);
ajStrAssignS(&seqid, ajSeqGetAccS(seq));
if(ajStrGetLen(seqid) == 0)
{
ajStrAssignS(&seqid, ajSeqGetNameS(seq));
}
if(ajStrGetLen(seqid) == 0)
{
ajWarn("No valid header information\n");
}
url = ajStrNew();
ajFmtPrintS(&url, "http://%S/%S/gcskew/window=%d/slide=%d/cumulative=%d/"
"at=%d/purine=%d/keto=%d/output=f/", base, restid, window,
slide, cumulative, at, purine, keto);
if(plot)
{
title = ajStrNew();
ajStrAppendC(&title, argv[0]);
ajStrAppendC(&title, " of ");
ajStrAppendS(&title, seqid);
gpp.title = ajStrNewS(title);
gpp.xlab = ajStrNewC("location");
gpp.ylab = ajStrNewC("GC skew");
if(!gFilebuffURLS(url, &buff))
{
ajDie("File downloading error from:\n%S\n", url);
}
if(!gPlotFilebuff(buff, mult, &gpp))
{
ajDie("Error in plotting\n");
}
AJFREE(gpp.title);
AJFREE(gpp.xlab);
AJFREE(gpp.ylab);
ajStrDel(&title);
ajFilebuffDel(&buff);
}
else
{
ajFmtPrintF(outf, "Sequence: %S\n", seqid);
if(!gFileOutURLS(url, &outf))
{
ajDie("File downloading error from:\n%S\n", url);
}
}
ajStrDel(&url);
ajStrDel(&restid);
ajStrDel(&seqid);
}
ajFileClose(&outf);
ajSeqallDel(&seqall);
ajSeqDel(&seq);
ajStrDel(&base);
embExit();
return 0;
}
static void jaspscan_scan(const AjPStr seq, const ajuint begin,
const AjPStr mfname, const char type,
const float threshold,
const AjBool both, AjPList hits)
{
PJsphits val = NULL;
AjPStr mname = NULL;
float **matrix = NULL;
ajuint cols;
ajuint i;
ajuint rc;
ajuint cc;
ajuint limit;
ajuint slen;
const char *p;
char schar;
float sum = 0.;
float rmax = 0.;
float scorepc = 0.;
float maxscore = 0.;
cols = jaspscan_readmatrix(mfname, &matrix);
maxscore = 0.;
for(cc = 0; cc < cols; ++cc)
{
rmax = 0.;
for(rc = 0; rc < 4; ++rc)
rmax = (rmax > matrix[rc][cc]) ? rmax : matrix[rc][cc];
maxscore += rmax;
}
slen = ajStrGetLen(seq);
limit = (slen - cols) + 1;
if(limit <= 0)
return;
mname = ajStrNew();
ajStrAssignS(&mname,mfname);
ajFilenameTrimPath(&mname);
ajFilenameTrimExt(&mname);
p = ajStrGetPtr(seq);
for(i=0; i < limit; ++i)
{
sum = 0.;
schar = p[i];
for(cc = 0; cc < cols; ++cc)
{
schar = p[i+cc];
if(schar == 'A')
sum += matrix[0][cc];
else if(schar == 'C')
sum += matrix[1][cc];
else if(schar == 'G')
sum += matrix[2][cc];
else if(schar == 'T')
sum += matrix[3][cc];
}
scorepc = (sum * (float)100.) / maxscore;
if(scorepc >= threshold)
{
val = jaspscan_hitsnew();
val->type = type;
ajStrAssignS(&val->matname,mname);
val->start = i + begin;
val->end = val->start + cols - 1;
val->score = sum;
val->threshold = threshold;
val->scorepc = scorepc;
val->maxscore = maxscore;
ajListPushAppend(hits,(void *)val);
}
}
if(both)
{
jaspscan_CompMat(matrix, cols);
p = ajStrGetPtr(seq);
for(i=0; i < limit; ++i)
{
sum = 0.;
schar = p[i];
for(cc = 0; cc < cols; ++cc)
{
schar = p[i+cc];
if(schar == 'A')
sum += matrix[0][cc];
else if(schar == 'C')
sum += matrix[1][cc];
else if(schar == 'G')
sum += matrix[2][cc];
else if(schar == 'T')
sum += matrix[3][cc];
}
scorepc = (sum * (float)100.) / maxscore;
if(scorepc >= threshold)
{
val = jaspscan_hitsnew();
val->type = type;
ajStrAssignS(&val->matname,mname);
val->end = i + begin;
val->start = val->end + cols - 1;
val->score = sum;
val->threshold = threshold;
val->scorepc = scorepc;
val->maxscore = maxscore;
ajListPushAppend(hits,(void *)val);
}
}
}
for(i = 0; i < 4; ++i)
AJFREE(matrix[i]);
AJFREE(matrix);
ajStrDel(&mname);
return;
}
static AjPTable jaspscan_ReadFamList(const AjPStr jaspdir)
{
AjPTable ret = NULL;
AjPStr lfile = NULL;
AjPStr line = NULL;
AjPStr key = NULL;
AjPStr str = NULL;
PJspmat info = NULL;
AjPFile inf = NULL;
const char *p = NULL;
const char *q = NULL;
lfile = ajStrNew();
line = ajStrNew();
str = ajStrNew();
ajFmtPrintS(&lfile,"%S%s%s",jaspdir,SLASH_STRING,J_LIST);
inf = ajDatafileNewInNameS(lfile);
if(!inf)
ajFatal("Matrix list file %S not found",lfile);
ret = ajTablestrNewLen(JASPTAB_GUESS);
while(ajReadlineTrim(inf,&line))
{
info = jaspscan_infonew();
ajFmtScanS(line,"%S%S",&info->id,&info->num);
p = ajStrGetPtr(line);
while(*p !='\t')
++p;
++p;
while(*p != '\t')
++p;
++p;
q = p;
while(*q != '\t')
++q;
ajStrAssignSubC(&info->name,p,0,q-p-1);
++q;
p = q;
while(*q != ';')
++q;
ajStrAssignSubC(&info->klass,p,0,q-p-1);
ajStrRemoveWhiteExcess(&info->klass);
p = q+1;
while(*p)
{
q = p;
while(* q && *q != ';')
++q;
ajStrAssignSubC(&str,p,0,q-p-1);
ajStrRemoveWhiteExcess(&str);
jaspscan_coretoken(info, str);
if(!*q)
p = q;
else
p = q +1;
}
key = ajStrNew();
ajStrAssignS(&key,info->id);
ajTablePut(ret,(void *)key,(void *) info);
}
ajFileClose(&inf);
ajStrDel(&lfile);
ajStrDel(&line);
ajStrDel(&str);
return ret;
}
int main(int argc, char *argv[])
{
embInitPV("gdeltagcskew", argc, argv, "GEMBASSY", "1.0.3");
AjPSeqall seqall;
AjPSeq seq;
AjPStr inseq = NULL;
AjBool at = 0;
AjBool purine = 0;
AjBool keto = 0;
AjPStr method = 0;
AjBool accid = ajFalse;
AjPStr restid = NULL;
AjPStr seqid = NULL;
AjPStr base = NULL;
AjPStr url = NULL;
AjPFile tmpfile = NULL;
AjPStr tmpname = NULL;
AjPFilebuff tmp = NULL;
AjPStr line = NULL;
AjPFile outf = NULL;
seqall = ajAcdGetSeqall("sequence");
at = ajAcdGetBoolean("at");
purine = ajAcdGetBoolean("purine");
keto = ajAcdGetBoolean("keto");
method = ajAcdGetSelectSingle("method");
accid = ajAcdGetBoolean("accid");
outf = ajAcdGetOutfile("outfile");
base = ajStrNewC("rest.g-language.org");
gAssignUniqueName(&tmpname);
while(ajSeqallNext(seqall, &seq))
{
inseq = NULL;
if(!accid)
{
if(gFormatGenbank(seq, &inseq))
{
tmpfile = ajFileNewOutNameS(tmpname);
if(!tmpfile)
{
ajDie("Output file (%S) open error\n", tmpname);
}
ajFmtPrintF(tmpfile, "%S", inseq);
ajFileClose(&tmpfile);
ajFmtPrintS(&url, "http://%S/upload/upl.pl", base);
gFilePostSS(url, tmpname, &restid);
ajStrDel(&url);
ajSysFileUnlinkS(tmpname);
}
else
{
ajWarn("Sequence does not have features\n"
"Proceeding with sequence accession ID\n");
accid = ajTrue;
}
}
ajStrAssignS(&seqid, ajSeqGetAccS(seq));
if(ajStrGetLen(seqid) == 0)
{
ajStrAssignS(&seqid, ajSeqGetNameS(seq));
}
if(ajStrGetLen(seqid) == 0)
{
ajWarn("No valid header information\n");
}
if(accid)
{
ajStrAssignS(&restid, seqid);
if(ajStrGetLen(seqid) == 0)
{
ajDie("Cannot proceed without header with -accid\n");
}
if(!gValID(seqid))
{
ajDie("Invalid accession ID:%S, exiting\n", seqid);
}
}
url = ajStrNew();
ajFmtPrintS(&url, "http://%S/%S/delta_gcskew/", base, restid);
if(!gFilebuffURLS(url, &tmp))
{
ajDie("Failed to download result from:\n%S\n", url);
}
ajBuffreadLine(tmp, &line);
ajStrRemoveSetC(&line, "\n");
ajFmtPrintF(outf, "Sequence: %S DELTA-GCskew %S\n", seqid, line);
ajStrDel(&url);
ajStrDel(&restid);
ajStrDel(&seqid);
ajStrDel(&inseq);
}
ajFileClose(&outf);
ajSeqallDel(&seqall);
ajSeqDel(&seq);
ajStrDel(&base);
embExit();
return 0;
}
