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;
}
int main(int argc, char *argv[])
{
embInitPV("genret", argc, argv, "GEMBASSY", "1.0.3");
AjPSeqall seqall;
AjPSeq seq = NULL;
AjPStr inseq = NULL;
AjPStr gene = NULL;
AjPStr access = NULL;
AjBool accid = ajTrue;
AjPStr argument = NULL;
AjPFile outfile = NULL;
AjPStr seqid = NULL;
AjPStr restid = NULL;
AjBool valid = ajFalse;
AjBool isseq = ajFalse;
AjBool isgbk = ajFalse;
AjPFilebuff buff = NULL;
AjPFile tmpfile = NULL;
AjPStr tmpname = NULL;
AjPStr regexstr = NULL;
AjPStrTok token = NULL;
AjPRegexp regex = NULL;
AjPStr url = NULL;
AjPStr base = NULL;
AjPStr head = NULL;
AjPStr line = NULL;
seqall = ajAcdGetSeqall("sequence");
access = ajAcdGetString("access");
gene = ajAcdGetString("gene");
argument = ajAcdGetString("argument");
accid = ajAcdGetBoolean("accid");
outfile = ajAcdGetOutfile("outfile");
if(
ajStrMatchC(access, "translation") ||
ajStrMatchC(access, "get_exon") ||
ajStrMatchC(access, "get_exons") ||
ajStrMatchC(access, "get_cdsseq") ||
ajStrMatchC(access, "get_gbkseq") ||
ajStrMatchC(access, "get_geneseq") ||
ajStrMatchC(access, "get_intron") ||
ajStrMatchC(access, "getseq") ||
ajStrMatchC(access, "seq") ||
ajStrMatchC(access, "around_startcodon") ||
ajStrMatchC(access, "around_stopcodon") ||
ajStrMatchC(access, "before_startcodon") ||
ajStrMatchC(access, "before_stopcodon") ||
ajStrMatchC(access, "after_startcodon") ||
ajStrMatchC(access, "after_stopcodon")
)
{
isseq = ajTrue;
}
else if(ajStrMatchC(access, "annotate") ||
ajStrMatchC(access, "output"))
{
isgbk = ajTrue;
}
else
{
ajFmtPrintF(outfile, "gene,%S\n", access);
}
base = ajStrNewC("rest.g-language.org");
ajStrExchangeCC(&argument, " ", "/");
ajStrExchangeCC(&argument, ",", "/");
ajStrExchangeCC(&argument, "\t", "/");
ajStrExchangeCC(&argument, "\r", "/");
ajStrExchangeCC(&argument, "\n", "/");
if(ajStrMatchC(gene, "*"))
{
ajStrInsertK(&gene, 0, '.');
}
if(ajStrPrefixC(gene, "@") || ajStrPrefixC(gene, "list::"))
{
ajStrExchangeCC(&gene, "@", "");
ajStrExchangeCC(&gene, "list::", "");
ajStrAssignS(&tmpname, gene);
tmpfile = ajFileNewInNameS(tmpname);
if(!tmpfile)
{
ajDie("List file (%S) open error\n", tmpname);
}
gene = ajStrNew();
while(ajReadline(tmpfile, &line))
{
ajStrAppendS(&gene, line);
}
ajFileClose(&tmpfile);
ajStrDel(&tmpname);
ajStrDel(&line);
}
tmpname = ajStrNew();
gAssignUniqueName(&tmpname);
while(ajSeqallNext(seqall, &seq))
{
inseq = ajStrNew();
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();
if(isgbk)
{
ajFmtPrintS(&url, "http://%S/%S/%S", base, restid, access);
}
else
{
ajFmtPrintS(&url, "http://%S/%S/*/%S/%S", base, restid, access, argument);
}
if(!gFilebuffURLS(url, &buff))
{
ajDie("GET error from %S\n", url);
}
while(ajBuffreadLine(buff, &line))
{
if(isgbk){
ajFmtPrintF(outfile, "%S", line);
continue;
}
ajStrRemoveLastNewline(&line);
regex = ajRegCompC("^>");
if(ajRegExec(regex, line))
{
head = ajStrNew();
ajStrAssignS(&head, line);
ajStrTrimStartC(&head, ">");
valid = ajFalse;
token = ajStrTokenNewC(ajStrNewS(gene), " ,\t\r\n");
while(ajStrTokenNextParse(token, ®exstr))
{
if(ajStrGetLen(regexstr))
{
regex = ajRegComp(regexstr);
if(ajRegExec(regex, line))
{
valid = ajTrue;
if(ajStrIsAlnum(regexstr))
{
ajStrExchangeSC(&gene, regexstr, "");
}
}
ajRegFree(®ex);
}
}
}
else
{
if(valid)
{
if(isseq)
{
ajStrFmtWrap(&line, 60);
ajFmtPrintF(outfile, ">%S\n%S\n", head, line);
}
else
{
ajFmtPrintF(outfile, "%S,%S\n", head, line);
}
valid = ajFalse;
}
}
}
ajFileClose(&outfile);
ajStrDel(&restid);
ajStrDel(&seqid);
ajStrDel(&inseq);
}
ajSeqallDel(&seqall);
ajSeqDel(&seq);
ajStrDel(&access);
ajStrDel(&gene);
embExit();
}
