static void jaspextract_getjaspdirs(AjPList jdirlist)
{
const AjPStr datadir = NULL;
AjPStr line = NULL;
ajuint n = 0;
ajuint i = 0;
datadir = ajDatafileValuePath();
if(!datadir)
ajFatal("jaspextract: Cannot determine the EMBOSS data directory");
n = ajFilelistAddPathDir(jdirlist, datadir);
for(i=0; i < n; ++i)
{
ajListPop(jdirlist,(void **)&line);
if(ajStrFindC(line,"JASPAR_") != -1)
ajListPushAppend(jdirlist,(void *)line);
else
ajStrDel(&line);
}
return;
}
static void remap_read_file_of_enzyme_names(AjPStr *enzymes)
{
AjPFile file = NULL;
AjPStr line;
const char *p = NULL;
if(ajStrFindC(*enzymes, "@") == 0)
{
ajStrTrimC(enzymes, "@"); /* remove the @ */
file = ajFileNewInNameS(*enzymes);
if(file == NULL)
ajFatal("Cannot open the file of enzyme names: '%S'", enzymes);
/* blank off the enzyme file name and replace with the enzyme names */
ajStrSetClear(enzymes);
line = ajStrNew();
while(ajReadlineTrim(file, &line))
{
p = ajStrGetPtr(line);
if(!*p || *p == '#' || *p == '!')
continue;
ajStrAppendS(enzymes, line);
ajStrAppendC(enzymes, ",");
}
ajStrDel(&line);
ajFileClose(&file);
}
return;
}
static void makeprotseq_parse_pepstats(AjPList* list,AjPFile data)
{
AjPStr line = ajStrNew();
AjPStr ch;
AjPStr chcopy = NULL;
const AjPStr tok;
double value = 0.0;
ajint count = 0;
ajint i = 0;
ajDebug("Parsing pepstats file.\n");
/* skip the lines before residues */
while(ajReadline(data,&line))
{
if(ajStrFindC(line,"Residue") == 0)
break;
}
/* parse residue part */
while(ajReadlineTrim(data,&line))
{
if(ajStrParseCountC(line," \t") == 0)
break;
ch = ajStrNew();
tok = ajStrParseWhite(line);
ajStrAppendS(&ch,tok);
ajStrFmtLower(&ch);
for(count = 1; count < 5; count++)
tok = ajStrParseWhite(NULL);
ajStrToDouble(tok,&value);
count = (ajint) (value * 100) +
((value - (int) value ) >= 0.5 ? 1 : 0);
for(i=0; i<count; i++)
{
chcopy = ajStrNewS(ch);
ajListstrPush(*list,chcopy);
}
ajStrDel(&ch);
}
ajStrDel(&line);
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;
}
AjPPatlistRegex ajPatlistRegexRead (const AjPStr patspec,
const AjPStr patname,
const AjPStr fmt,
ajuint type, AjBool upper, AjBool lower)
{
AjPPatlistRegex patlist = NULL;
AjPStr line = NULL;
AjPStr pat = NULL;
AjPStr name = NULL;
AjPFilebuff infile = NULL;
AjPStr patstr = NULL;
ajuint ifmt;
ajuint npat = 0;
AjPStr namestr = NULL;
ajStrAssignS(&namestr, patname);
ajStrAssignEmptyC(&namestr, "regex");
ajStrAssignS(&patstr, patspec);
patlist = ajPatlistRegexNewType(type);
ifmt = patternRegexFormat(fmt);
if(ajStrGetCharFirst(patspec) == '@')
{
ajStrCutStart(&patstr, 1);
infile = ajFilebuffNewNameS(patstr);
if(!infile)
{
ajErr("Unable to open regular expression file '%S'", patstr);
return NULL;
}
line = ajStrNew();
pat = ajStrNew();
name = ajStrNew();
if(!ifmt)
{
ajBuffreadLineTrim(infile,&line);
if(ajStrPrefixC(line, ">"))
ifmt = 2;
else
ifmt = 1;
ajFilebuffReset(infile);
}
switch(ifmt)
{
case 1:
while (ajBuffreadLineTrim(infile,&line))
{
npat++;
ajStrAppendS (&pat,line);
if(lower)
ajStrFmtLower(&pat);
if(upper)
ajStrFmtUpper(&pat);
ajFmtPrintS(&name, "%S%u", namestr, npat);
ajPatternRegexNewList(patlist,name,pat);
ajStrSetClear(&pat);
}
break;
default:
while (ajBuffreadLineTrim(infile,&line))
{
if (ajStrFindC(line,">")>-1)
{
npat++;
if (ajStrGetLen(name))
{
if(lower)
ajStrFmtLower(&pat);
if(upper)
ajStrFmtUpper(&pat);
ajPatternRegexNewList(patlist,name,pat);
ajStrSetClear(&name);
ajStrSetClear(&pat);
}
ajStrCutStart(&line,1);
ajStrAssignS (&name,line);
if(!ajStrGetLen(name))
ajFmtPrintS(&name, "%S%u", namestr, npat);
}
else
ajStrAppendS (&pat,line);
}
ajStrAssignEmptyS(&name, patname);
ajPatternRegexNewList(patlist,name,pat);
ajStrSetClear(&pat);
break;
}
ajFilebuffDel(&infile);
}
else
{
ajStrAssignS(&pat, patspec);
if(lower)
ajStrFmtLower(&pat);
if(upper)
ajStrFmtUpper(&pat);
ajStrAssignS(&name, namestr);
ajPatternRegexNewList(patlist,name,pat);
}
ajStrDel(&name);
ajStrDel(&namestr);
ajStrDel(&patstr);
ajStrDel(&line);
ajStrDel(&pat);
return patlist;
}
int main(int argc, char **argv)
{
AjPSeqout outseq = NULL;
AjPList list = NULL;
AjPSeq seq = NULL;
AjPStr insert = NULL;
AjPStr seqstr = NULL;
AjPStr* seqr = NULL;
AjPFile data = NULL;
ajint start = 0;
ajint length = 0;
ajint amount = 0;
ajint scmax = 0;
ajint extra = 0;
embInit("makeprotseq", argc, argv);
data = ajAcdGetInfile("pepstatsfile");
insert = ajAcdGetString("insert");
start = ajAcdGetInt("start");
length = ajAcdGetInt("length");
amount = ajAcdGetInt("amount");
outseq = ajAcdGetSeqoutall("outseq");
list = ajListstrNew();
/* this is checked by acd
if(amount <=0 || length <= 0)
ajFatal("Amount or length is 0 or less. "
"Unable to create any sequences"); */
/* if insert, make sure sequence is large enough */
if(ajStrGetLen(insert))
{
length -= ajStrGetLen(insert);
/* start= start <= 1 ? 0 : --start; */ /* checked in acd */
start--;
if(length <= 0)
ajFatal("Sequence smaller than inserted part. "
"Unable to create sequences.");
}
/* make the list of AjPStr to be used in sequence creation */
if(data)
{
ajDebug("Distribution datafile '%s' given checking type\n",
ajFileGetPrintnameC(data));
seqstr = ajStrNew();
ajReadlineTrim(data,&seqstr);
if(ajStrFindC(seqstr,"PEPSTATS") == 0)
{
makeprotseq_parse_pepstats(&list,data);
}
else
{
ajWarn("Not pepstats file. Making completely random sequences.");
makeprotseq_default_chars(&list);
}
ajStrDel(&seqstr);
ajFileClose(&data);
}
else
makeprotseq_default_chars(&list);
/* if insert, make sure type is correct */
/* typecheking code is not working, uncomment and test after it is
if(ajStrGetLen(insert))
{
seqstr = ajStrNew();
if(prot)
ajStrAssignC(&seqstr,"pureprotein");
if(!ajSeqTypeCheckS(&insert,seqstr))
ajFatal("Insert not the same sequence type as sequence itself.");
ajStrDel(&seqstr);
} */
/* array allows fast creation of a sequences */
scmax = (ajuint) ajListstrToarray(list,&seqr);
if(!scmax)
ajFatal("No strings in list. No characters to make the sequence.");
ajDebug("Distribution array done.\nscmax '%d', extra '%d', first '%S'\n",
scmax,extra,seqr[0]);
ajRandomSeed();
while(amount-- > 0)
{
seqstr = makeprotseq_random_sequence(seqr,scmax,length);
if(ajStrGetLen(insert))
ajStrInsertS(&seqstr,start,insert);
ajStrFmtLower(&seqstr);
seq = ajSeqNew();
ajSeqAssignSeqS(seq, seqstr);
ajSeqSetProt(seq);
ajSeqoutWriteSeq(outseq, seq);
ajSeqDel(&seq);
ajStrDel(&seqstr);
}
ajSeqoutClose(outseq);
ajSeqoutDel(&outseq);
ajListstrFreeData(&list);
ajStrDel(&insert);
AJFREE(seqr);
embExit();
return 0;
}
