int main(int argc, char **argv)
{
AjPSeqall seqall = NULL;
AjPFile dend_outfile = NULL;
AjPStr tmp_dendfilename = NULL;
AjPFile tmp_dendfile = NULL;
AjPStr tmp_aln_outfile = NULL;
AjPSeqset seqset = NULL;
AjPSeqout seqout = NULL;
AjPSeqin seqin = NULL;
AjBool only_dend;
AjBool are_prot = ajFalse;
AjBool do_slow;
AjBool use_dend;
AjPFile dend_file = NULL;
AjPStr dend_filename = NULL;
ajint ktup;
ajint gapw;
ajint topdiags;
ajint window;
AjBool nopercent;
AjPStr pw_matrix = NULL;
AjPStr pw_dna_matrix = NULL;
AjPFile pairwise_matrix = NULL;
float pw_gapc;
float pw_gapv;
AjPStr pwmstr = NULL;
char pwmc = '\0';
AjPStr pwdstr = NULL;
char pwdc = '\0';
AjPStr m1str = NULL;
AjPStr m2str = NULL;
char m1c = '\0';
char m2c = '\0';
AjPStr matrix = NULL;
AjPStr dna_matrix = NULL;
AjPFile ma_matrix = NULL;
float gapc;
float gapv;
AjBool endgaps;
AjBool norgap;
AjBool nohgap;
ajint gap_dist;
ajint maxdiv;
AjPStr hgapres = NULL;
AjPSeqout fil_file = NULL;
AjPSeq seq = NULL;
AjPStr cmd = NULL;
AjPStr tmp = NULL;
AjPStr tmpFilename;
AjPStr line = NULL;
ajint nb = 0;
/* get all the parameters */
embInit("emma", argc, argv);
pwmstr = ajStrNew();
pwdstr = ajStrNew();
m1str = ajStrNew();
m2str = ajStrNew();
seqall = ajAcdGetSeqall("sequence");
seqout = ajAcdGetSeqoutset("outseq");
dend_outfile = ajAcdGetOutfile("dendoutfile");
only_dend = ajAcdGetToggle("onlydend");
use_dend = ajAcdGetToggle("dendreuse");
dend_file = ajAcdGetInfile("dendfile");
if (dend_file)
ajStrAssignS(&dend_filename, ajFileGetPrintnameS(dend_file));
ajFileClose(&dend_file);
do_slow = ajAcdGetToggle("slowalign");
ktup = ajAcdGetInt("ktup");
gapw = ajAcdGetInt("gapw");
topdiags = ajAcdGetInt("topdiags");
window = ajAcdGetInt("window");
nopercent = ajAcdGetBoolean("nopercent");
pw_matrix = ajAcdGetListSingle("pwmatrix");
pwmc = ajStrGetCharFirst(pw_matrix);
if(pwmc=='b')
ajStrAssignC(&pwmstr,"blosum");
else if(pwmc=='p')
ajStrAssignC(&pwmstr,"pam");
else if(pwmc=='g')
ajStrAssignC(&pwmstr,"gonnet");
else if(pwmc=='i')
ajStrAssignC(&pwmstr,"id");
else if(pwmc=='o')
ajStrAssignC(&pwmstr,"own");
pw_dna_matrix = ajAcdGetListSingle("pwdnamatrix");
pwdc = ajStrGetCharFirst(pw_dna_matrix);
if(pwdc=='i')
ajStrAssignC(&pwdstr,"iub");
else if(pwdc=='c')
ajStrAssignC(&pwdstr,"clustalw");
else if(pwdc=='o')
ajStrAssignC(&pwdstr,"own");
pairwise_matrix = ajAcdGetInfile("pairwisedatafile");
pw_gapc = ajAcdGetFloat( "pwgapopen");
pw_gapv = ajAcdGetFloat( "pwgapextend");
matrix = ajAcdGetListSingle( "matrix");
m1c = ajStrGetCharFirst(matrix);
if(m1c=='b')
ajStrAssignC(&m1str,"blosum");
else if(m1c=='p')
ajStrAssignC(&m1str,"pam");
else if(m1c=='g')
ajStrAssignC(&m1str,"gonnet");
else if(m1c=='i')
ajStrAssignC(&m1str,"id");
else if(m1c=='o')
ajStrAssignC(&m1str,"own");
dna_matrix = ajAcdGetListSingle( "dnamatrix");
m2c = ajStrGetCharFirst(dna_matrix);
if(m2c=='i')
ajStrAssignC(&m2str,"iub");
else if(m2c=='c')
ajStrAssignC(&m2str,"clustalw");
else if(m2c=='o')
ajStrAssignC(&m2str,"own");
ma_matrix = ajAcdGetInfile("mamatrixfile");
gapc = ajAcdGetFloat("gapopen");
gapv = ajAcdGetFloat("gapextend");
endgaps = ajAcdGetBoolean("endgaps");
norgap = ajAcdGetBoolean("norgap");
nohgap = ajAcdGetBoolean("nohgap");
gap_dist = ajAcdGetInt("gapdist");
hgapres = ajAcdGetString("hgapres");
maxdiv = ajAcdGetInt("maxdiv");
tmp = ajStrNewC("fasta");
/*
** Start by writing sequences into a unique temporary file
** get file pointer to unique file
*/
fil_file = ajSeqoutNew();
tmpFilename = emma_getUniqueFileName();
if(!ajSeqoutOpenFilename( fil_file, tmpFilename))
embExitBad();
/* Set output format to fasta */
ajSeqoutSetFormatS( fil_file, tmp);
while(ajSeqallNext(seqall, &seq))
{
/*
** Check sequences are all of the same type
** Still to be done
** Write out sequences
*/
if (!nb)
are_prot = ajSeqIsProt(seq);
ajSeqoutWriteSeq(fil_file, seq);
++nb;
}
ajSeqoutClose(fil_file);
if(nb < 2)
ajFatal("Multiple alignments need at least two sequences");
/* Generate clustalw command line */
cmd = ajStrNewS(ajAcdGetpathC("clustalw"));
/* add tmp file containing sequences */
ajStrAppendC(&cmd, " -infile=");
ajStrAppendS(&cmd, tmpFilename);
/* add out file name */
tmp_aln_outfile = emma_getUniqueFileName();
ajStrAppendC(&cmd, " -outfile=");
ajStrAppendS(&cmd, tmp_aln_outfile);
/* calculating just the nj tree or doing full alignment */
if(only_dend)
ajStrAppendC(&cmd, " -tree");
else
if(!use_dend)
ajStrAppendC(&cmd, " -align");
/* Set sequence type from information from acd file */
if(are_prot)
ajStrAppendC(&cmd, " -type=protein");
else
ajStrAppendC(&cmd, " -type=dna");
/*
** set output to MSF format - will read in this file later and output
** user requested format
*/
ajStrAppendC(&cmd, " -output=");
ajStrAppendC(&cmd, "gcg");
/* If going to do pairwise alignment */
if(!use_dend)
{
/* add fast pairwise alignments*/
if(!do_slow)
{
ajStrAppendC(&cmd, " -quicktree");
ajStrAppendC(&cmd, " -ktuple=");
ajStrFromInt(&tmp, ktup);
ajStrAppendS(&cmd, tmp);
ajStrAppendC(&cmd, " -window=");
ajStrFromInt(&tmp, window);
ajStrAppendS(&cmd, tmp);
if(nopercent)
ajStrAppendC(&cmd, " -score=percent");
else
ajStrAppendC(&cmd, " -score=absolute");
ajStrAppendC(&cmd, " -topdiags=");
ajStrFromInt(&tmp, topdiags);
ajStrAppendS(&cmd, tmp);
ajStrAppendC(&cmd, " -pairgap=");
ajStrFromInt(&tmp, gapw);
ajStrAppendS(&cmd, tmp);
}
else
{
if(pairwise_matrix)
{
if(are_prot)
ajStrAppendC(&cmd, " -pwmatrix=");
else
ajStrAppendC(&cmd, " -pwdnamatrix=");
ajStrAppendS(&cmd, ajFileGetPrintnameS(pairwise_matrix));
}
else
{
if(are_prot)
{
ajStrAppendC(&cmd, " -pwmatrix=");
ajStrAppendS(&cmd, pwmstr);
}
else
{
ajStrAppendC(&cmd, " -pwdnamatrix=");
ajStrAppendS(&cmd, pwdstr);
}
}
ajStrAppendC(&cmd, " -pwgapopen=");
ajStrFromFloat(&tmp, pw_gapc, 3);
ajStrAppendS(&cmd, tmp);
ajStrAppendC(&cmd, " -pwgapext=");
ajStrFromFloat(&tmp, pw_gapv, 3);
ajStrAppendS(&cmd, tmp);
}
}
/* Multiple alignments */
/* using existing tree or generating new tree? */
if(use_dend)
{
ajStrAppendC(&cmd, " -usetree=");
ajStrAppendS(&cmd, dend_filename);
}
else
{
/* use tmp file to hold dend file, will read back in later */
tmp_dendfilename = emma_getUniqueFileName();
ajStrAppendC(&cmd, " -newtree=");
ajStrAppendS(&cmd, tmp_dendfilename);
}
if(ma_matrix)
{
if(are_prot)
ajStrAppendC(&cmd, " -matrix=");
else
ajStrAppendC(&cmd, " -pwmatrix=");
ajStrAppendS(&cmd, ajFileGetPrintnameS(ma_matrix));
}
else
{
if(are_prot)
{
ajStrAppendC(&cmd, " -matrix=");
ajStrAppendS(&cmd, m1str);
}
else
{
ajStrAppendC(&cmd, " -dnamatrix=");
ajStrAppendS(&cmd, m2str);
}
}
ajStrAppendC(&cmd, " -gapopen=");
ajStrFromFloat(&tmp, gapc, 3);
ajStrAppendS(&cmd, tmp);
ajStrAppendC(&cmd, " -gapext=");
ajStrFromFloat(&tmp, gapv, 3);
ajStrAppendS(&cmd, tmp);
ajStrAppendC(&cmd, " -gapdist=");
ajStrFromInt(&tmp, gap_dist);
ajStrAppendS(&cmd, tmp);
ajStrAppendC(&cmd, " -hgapresidues=");
ajStrAppendS(&cmd, hgapres);
if(!endgaps)
ajStrAppendC(&cmd, " -endgaps");
if(norgap)
ajStrAppendC(&cmd, " -nopgap");
if(nohgap)
ajStrAppendC(&cmd, " -nohgap");
ajStrAppendC(&cmd, " -maxdiv=");
ajStrFromInt(&tmp, maxdiv);
ajStrAppendS(&cmd, tmp);
/* run clustalw */
/* ajFmtError("..%s..\n\n", ajStrGetPtr( cmd)); */
ajDebug("Executing '%S'\n", cmd);
ajSysExecS(cmd);
/* produce alignment file only if one was produced */
if(!only_dend)
{
/* read in tmp alignment output file to output through EMBOSS output */
seqin = ajSeqinNew();
/*
** add the Usa format to the start of the filename to tell EMBOSS
** format of file
*/
ajStrInsertC(&tmp_aln_outfile, 0, "msf::");
ajSeqinUsa(&seqin, tmp_aln_outfile);
seqset = ajSeqsetNew();
if(ajSeqsetRead(seqset, seqin))
{
ajSeqoutWriteSet(seqout, seqset);
ajSeqoutClose(seqout);
ajSeqinDel(&seqin);
/* remove the Usa from the start of the string */
ajStrCutStart(&tmp_aln_outfile, 5);
}
else
ajFmtError("Problem writing out EMBOSS alignment file\n");
}
/* read in new tmp dend file (if produced) to output through EMBOSS */
if(tmp_dendfilename!=NULL)
{
tmp_dendfile = ajFileNewInNameS( tmp_dendfilename);
if(tmp_dendfile!=NULL){
while(ajReadlineTrim(tmp_dendfile, &line))
ajFmtPrintF(dend_outfile, "%s\n", ajStrGetPtr( line));
ajFileClose(&tmp_dendfile);
ajSysFileUnlinkS(tmp_dendfilename);
}
}
ajSysFileUnlinkS(tmpFilename);
if(!only_dend)
ajSysFileUnlinkS(tmp_aln_outfile);
ajStrDel(&pw_matrix);
ajStrDel(&matrix);
ajStrDel(&pw_dna_matrix);
ajStrDel(&dna_matrix);
ajStrDel(&tmp_dendfilename);
ajStrDel(&dend_filename);
ajStrDel(&tmp_aln_outfile);
ajStrDel(&pwmstr);
ajStrDel(&pwdstr);
ajStrDel(&m1str);
ajStrDel(&m2str);
ajStrDel(&hgapres);
ajStrDel(&cmd);
ajStrDel(&tmp);
ajStrDel(&tmpFilename);
ajStrDel(&line);
ajFileClose(&dend_outfile);
ajFileClose(&tmp_dendfile);
ajFileClose(&dend_file);
ajFileClose(&pairwise_matrix);
ajFileClose(&ma_matrix);
ajSeqallDel(&seqall);
ajSeqsetDel(&seqset);
ajSeqDel(&seq);
ajSeqoutDel(&seqout);
ajSeqoutDel(&fil_file);
ajSeqinDel(&seqin);
embExit();
return 0;
}
int main(int argc, char **argv)
{
AjPSeqset seqset = NULL;
AjPStr cl = NULL;
AjPSeqout seqout = NULL;
AjBool full = ajFalse;
AjPStr fn = NULL;
AjPStr stmp = NULL;
AjPStr outfname = NULL;
embInitPV("echlorop", argc, argv, "CBSTOOLS", VERSION);
seqset = ajAcdGetSeqset("sequence");
outfname= ajAcdGetOutfileName("outfile");
full = ajAcdGetBoolean("full");
cl = ajStrNewS(ajAcdGetpathC("chlorop"));
fn = ajStrNew();
stmp = ajStrNew();
ajFilenameSetTempname(&fn);
seqout = ajSeqoutNew();
if(!ajSeqoutOpenFilename(seqout, fn))
ajFatal("Cannot open temporary file %S",fn);
ajSeqoutSetFormatC(seqout, "fasta");
ajSeqoutWriteSet(seqout,seqset);
ajSeqoutClose(seqout);
if(full)
ajStrAppendC(&cl," -F");
ajFmtPrintS(&stmp," %S",fn);
ajStrAppendS(&cl,stmp);
#if 0
ajFmtPrint("%S\n",cl);
#endif
#if 1
ajSysExecOutnameAppendS(cl, outfname);
#endif
ajSysFileUnlinkS(fn);
ajStrDel(&cl);
ajStrDel(&fn);
ajStrDel(&stmp);
ajSeqoutDel(&seqout);
ajSeqsetDel(&seqset);
ajStrDel(&outfname);
embExit();
return 0;
}
int main(int argc, char **argv)
{
AjPSeqall nucseq; /* input nucleic sequences */
AjPSeqset protseq; /* input aligned protein sequences */
AjPSeqout seqout;
AjPSeq nseq; /* next nucleic sequence to align */
const AjPSeq pseq; /* next protein sequence use in alignment */
AjPTrn trnTable;
AjPSeq pep; /* translation of nseq */
AjPStr tablelist;
ajint table;
AjPSeqset outseqset; /* set of aligned nucleic sequences */
ajint proteinseqcount = 0;
AjPStr degapstr = NULL;
/* used to check if it matches with START removed */
AjPStr degapstr2 = NULL;
AjPStr codon = NULL; /* holds temporary codon to check if is START */
char aa; /* translated putative START codon */
ajint type; /* returned type of the putative START codon */
/* start position of guide protein in translation */
ajlong pos = 0;
AjPSeq newseq = NULL; /* output aligned nucleic sequence */
ajint frame;
embInit("tranalign", argc, argv);
nucseq = ajAcdGetSeqall("asequence");
protseq = ajAcdGetSeqset("bsequence");
tablelist = ajAcdGetListSingle("table");
seqout = ajAcdGetSeqoutset("outseq");
outseqset = ajSeqsetNew();
degapstr = ajStrNew();
/* initialise the translation table */
ajStrToInt(tablelist, &table);
trnTable = ajTrnNewI(table);
ajSeqsetFill(protseq);
while(ajSeqallNext(nucseq, &nseq))
{
if((pseq = ajSeqsetGetseqSeq(protseq, proteinseqcount++)) == NULL)
ajErr("No guide protein sequence available for "
"nucleic sequence %S",
ajSeqGetNameS(nseq));
ajDebug("Aligning %S and %S\n",
ajSeqGetNameS(nseq), ajSeqGetNameS(pseq));
/* get copy of pseq string with no gaps */
ajStrAssignS(°apstr, ajSeqGetSeqS(pseq));
ajStrRemoveGap(°apstr);
/*
** for each translation frame look for subset of pep that
** matches pseq
*/
for(frame = 1; frame <4; frame++)
{
ajDebug("trying frame %d\n", frame);
pep = ajTrnSeqOrig(trnTable, nseq, frame);
degapstr2 = ajStrNew();
ajStrAssignRef(°apstr2, degapstr);
pos = ajStrFindCaseS(ajSeqGetSeqS(pep), degapstr);
/*
** we might have a START codon that should be translated as 'M'
** we need to check if there is a match after a possible START
** codon
*/
if(pos == -1 && ajStrGetLen(degapstr) > 1 &&
(ajStrGetPtr(degapstr)[0] == 'M' ||
ajStrGetPtr(degapstr)[0] == 'm'))
{
/* see if pep minus the first character is a match */
ajStrCutStart(°apstr2, 1);
pos = ajStrFindCaseS(ajSeqGetSeqS(pep), degapstr2);
/*
** pos is >= 1 if we have a match that is after the first
** residue
*/
if(pos >= 1)
{
/* point back at the putative START Methionine */
pos--;
/* test if first codon is a START */
codon = ajStrNew();
ajStrAssignSubS(&codon, ajSeqGetSeqS(nseq),
(pos*3)+frame-1, (pos*3)+frame+2);
type = ajTrnCodonstrTypeS(trnTable, codon, &aa);
if(type != 1)
{
/* first codon is not a valid START, force a mismatch */
pos = -1;
}
ajStrDel(&codon);
}
else
{
/* force 'pos == 0' to be treated as a mismatch */
pos = -1;
}
}
ajStrDel(°apstr2);
ajSeqDel(&pep);
if(pos != -1)
break;
}
if(pos == -1)
ajErr("Guide protein sequence %S not found in nucleic sequence %S",
ajSeqGetNameS(pseq), ajSeqGetNameS(nseq));
else
{
ajDebug("got a match with frame=%d\n", frame);
/* extract the coding region of nseq with gaps */
newseq = ajSeqNew();
ajSeqSetNuc(newseq);
ajSeqAssignNameS(newseq, ajSeqGetNameS(nseq));
ajSeqAssignDescS(newseq, ajSeqGetDescS(nseq));
tranalign_AddGaps(newseq, nseq, pseq, (pos*3)+frame-1);
/* output the gapped nucleic sequence */
ajSeqsetApp(outseqset, newseq);
ajSeqDel(&newseq);
}
ajStrRemoveWhiteExcess(°apstr);
}
ajSeqoutWriteSet(seqout, outseqset);
ajSeqoutClose(seqout);
ajTrnDel(&trnTable);
ajSeqsetDel(&outseqset);
ajStrDel(°apstr);
ajStrDel(°apstr2);
ajSeqallDel(&nucseq);
ajSeqDel(&nseq);
ajSeqoutDel(&seqout);
ajSeqsetDel(&protseq);
ajStrDel(&tablelist);
embExit();
return 0;
}
int main(int argc, char **argv)
{
AjPSeqset seqset = NULL;
AjPStr cl = NULL;
AjPSeqout seqout = NULL;
float cutoff = 0.;
AjBool best = ajFalse;
AjBool gff = ajFalse;
AjBool two = ajFalse;
AjBool kinase = ajFalse;
AjPStr rsd = NULL;
AjBool addseq = ajFalse;
AjBool plot = ajFalse;
const AjPStr ofn = NULL;
AjPStr fn = NULL;
AjPStr stmp = NULL;
AjPFile outf = NULL;
embInitPV("enetphos", argc, argv, "CBSTOOLS", VERSION);
seqset = ajAcdGetSeqset("sequence");
outf = ajAcdGetOutfile("outfile");
plot = ajAcdGetBoolean("plot");
best = ajAcdGetBoolean("best");
gff = ajAcdGetBoolean("gff");
two = ajAcdGetBoolean("two");
kinase = ajAcdGetBoolean("kinase");
addseq = ajAcdGetBoolean("addseq");
cutoff = ajAcdGetFloat("cutoff");
rsd = ajAcdGetListSingle("residue");
cl = ajStrNewC("netphos ");
fn = ajStrNew();
stmp = ajStrNew();
ajFilenameSetTempname(&fn);
seqout = ajSeqoutNew();
if(!ajSeqoutOpenFilename(seqout, fn))
ajFatal("Cannot open temporary file %S",fn);
ajSeqoutSetFormatC(seqout, "fasta");
ajSeqoutWriteSet(seqout,seqset);
ajSeqoutClose(seqout);
if(ajStrMatchC(rsd,"serine"))
ajStrAppendC(&cl," -s");
else if(ajStrMatchC(rsd,"threonine"))
ajStrAppendC(&cl," -t");
else if(ajStrMatchC(rsd,"tyrosine"))
ajStrAppendC(&cl," -y");
if(plot)
ajStrAppendC(&cl," -g");
if(two)
ajStrAppendC(&cl," -2");
if(best)
ajStrAppendC(&cl," -b");
if(gff)
ajStrAppendC(&cl," -f gff");
if(kinase)
ajStrAppendC(&cl," -k");
if(addseq)
ajStrAppendC(&cl," -S");
ajFmtPrintS(&stmp," -c %f",cutoff);
ajStrAppendS(&cl,stmp);
ajFmtPrintS(&stmp," %S",fn);
ajStrAppendS(&cl,stmp);
ofn = ajFileGetNameS(outf);
ajFmtPrintS(&stmp," > %S",ofn);
ajStrAppendS(&cl,stmp);
ajFileClose(&outf);
#if 0
ajFmtPrint("%S\n",cl);
#endif
#if 1
if(system(ajStrGetPtr(cl)) == -1)
ajFatal("Command %S failed",cl);
#endif
ajSysFileUnlink(fn);
ajStrDel(&cl);
ajStrDel(&fn);
ajStrDel(&stmp);
ajStrDel(&rsd);
ajSeqoutDel(&seqout);
ajSeqsetDel(&seqset);
embExit();
return 0;
}
int main(int argc, char **argv)
{
/* ACD data item variables */
AjPSeqset dataset = NULL;
AjPFile bfile = NULL;
AjPFile plib = NULL;
AjPStr mod = NULL;
ajint nmotifs = 0;
AjBool text = ajFalse;
AjPStr prior = NULL;
float evt = 0.0;
ajint nsites = 0;
ajint minsites = 0;
ajint maxsites = 0;
float wnsites = 0.0;
ajint w = 0;
ajint minw = 0;
ajint maxw = 0;
AjBool nomatrim = ajFalse;
ajint wg = 0;
ajint ws = 0;
AjBool noendgaps = ajFalse;
AjBool revcomp = ajFalse;
AjBool pal = ajFalse;
AjBool nostatus = ajFalse;
ajint maxiter = 0;
float distance = 0.0;
float b = 0.0;
float spfuzz = 0.0;
AjPStr spmap = NULL;
AjPStr cons = NULL;
ajint maxsize = 0;
ajint p = 0;
ajint time = 0;
AjPStr sf = NULL;
ajint heapsize = 64;
AjBool xbranch = ajFalse;
AjBool wbranch = ajFalse;
ajint bfactor = 0;
AjPFile outtext = NULL;
/* Housekeeping variables */
AjPStr cmd = NULL;
AjPStr ssname = NULL;
AjPSeqout outseq = NULL;
AjPStr tmp = NULL;
char option;
/* ACD file processing */
embInitPV("ememetext",argc,argv,"MEME",VERSION);
dataset = ajAcdGetSeqset("dataset");
bfile = ajAcdGetInfile("bfile");
plib = ajAcdGetInfile("plibfile");
mod = ajAcdGetSelectSingle("mod");
nmotifs = ajAcdGetInt("nmotifs");
text = ajAcdGetBoolean("text");
prior = ajAcdGetSelectSingle("prior");
evt = ajAcdGetFloat("evt");
nsites = ajAcdGetInt("nsites");
minsites = ajAcdGetInt("minsites");
maxsites = ajAcdGetInt("maxsites");
wnsites = ajAcdGetFloat("wnsites");
w = ajAcdGetInt("w");
minw = ajAcdGetInt("minw");
maxw = ajAcdGetInt("maxw");
nomatrim = ajAcdGetBoolean("nomatrim");
wg = ajAcdGetInt("wg");
ws = ajAcdGetInt("ws");
noendgaps = ajAcdGetBoolean("noendgaps");
revcomp = ajAcdGetBoolean("revcomp");
pal = ajAcdGetBoolean("pal");
nostatus = ajAcdGetBoolean("nostatus");
maxiter = ajAcdGetInt("maxiter");
distance = ajAcdGetFloat("distance");
b = ajAcdGetFloat("b");
spfuzz = ajAcdGetFloat("spfuzz");
spmap = ajAcdGetSelectSingle("spmap");
cons = ajAcdGetString("cons");
maxsize = ajAcdGetInt("maxsize");
p = ajAcdGetInt("p");
time = ajAcdGetInt("time");
sf = ajAcdGetString("sf");
heapsize = ajAcdGetInt("heapsize");
xbranch = ajAcdGetBoolean("xbranch");
wbranch = ajAcdGetBoolean("wbranch");
bfactor = ajAcdGetInt("bfactor");
outtext = ajAcdGetOutfile("outtext");
outseq = ajAcdGetSeqoutset("outseq");
/* MAIN APPLICATION CODE */
/* 1. Housekeeping */
cmd = ajStrNew();
tmp = ajStrNew();
/* 2. Re-write dataset to a temporary file in a format (fasta) MEME
** can understand.
** Can't just pass the name of dataset to MEME as the name provided
** might be a USA which MEME would not understand.
*/
ssname = ajStrNewS(ajFileGetNameS(outseq->File));
ajSeqoutSetFormatC(outseq, "fasta");
ajSeqoutWriteSet(outseq, dataset);
ajSeqoutClose(outseq);
ajSeqoutDel(&outseq);
/* 3. Build ememe command line */
/* Command line is built in this order:
i. Application name.
ii. Original MEME options (in order they appear in ACD file)
iii.Original MEME options (that don't appear in ACD file)
iv. EMBASSY MEME new qualifiers and parameters.
*/
ajStrAssignS(&cmd, ajAcdGetpathC("meme"));
ajFmtPrintAppS(&cmd, " %S", ssname);
if(bfile)
ajFmtPrintAppS(&cmd, " -bfile %s ", ajFileGetNameC(bfile));
if(plib)
ajFmtPrintAppS(&cmd, " -plib %s ", ajFileGetNameC(plib));
option = ajStrGetCharFirst(mod);
if(option == 'o')
ajStrAppendC(&cmd, " -mod oops ");
else if(option == 'z')
ajStrAppendC(&cmd, " -mod zoops ");
else if(option == 'a')
ajStrAppendC(&cmd, " -mod anr ");
if(nmotifs != 1)
ajFmtPrintAppS(&cmd, " -nmotifs %d ", nmotifs);
if(text)
ajFmtPrintAppS(&cmd, " -text ");
ajFmtPrintAppS(&cmd, " -prior %S ", prior);
if(evt != -1)
ajFmtPrintAppS(&cmd, " -evt %f ", evt);
if(nsites != -1)
ajFmtPrintAppS(&cmd, " -nsites %d ", nsites);
else
{
if(minsites != -1)
ajFmtPrintAppS(&cmd, " -minsites %d ", minsites);
if(maxsites != -1)
ajFmtPrintAppS(&cmd, " -maxsites %d ", maxsites);
}
if(wnsites < 0.7999 || wnsites > .8001)
ajFmtPrintAppS(&cmd, " -wnsites %f ", wnsites);
if(w != -1)
ajFmtPrintAppS(&cmd, " -w %d ", w);
if(minw != 8)
ajFmtPrintAppS(&cmd, " -minw %d ", minw);
if(maxw != 50)
ajFmtPrintAppS(&cmd, " -maxw %d ", maxw);
if(nomatrim)
ajFmtPrintAppS(&cmd, " -nomatrim ");
if(wg != 11)
ajFmtPrintAppS(&cmd, " -wg %d ", wg);
if(ws != 1)
ajFmtPrintAppS(&cmd, " -ws %d ", ws);
if(noendgaps)
ajFmtPrintAppS(&cmd, " -noendgaps ");
if(revcomp)
ajFmtPrintAppS(&cmd, " -revcomp ");
if(pal && ajSeqsetIsNuc(dataset))
ajFmtPrintAppS(&cmd, " -pal ");
if(nostatus)
ajFmtPrintAppS(&cmd, " -nostatus ");
if(maxiter != 50)
ajFmtPrintAppS(&cmd, " -maxiter %d ", maxiter);
if(distance < 0.00099 || distance > 0.00101)
ajFmtPrintAppS(&cmd, " -distance %f ", distance);
if(b != -1)
ajFmtPrintAppS(&cmd, " -b %f ", b);
if(spfuzz != -1)
ajFmtPrintAppS(&cmd, " -spfuzz %f ", spfuzz);
if(!ajStrMatchC(spmap,"default"))
ajFmtPrintAppS(&cmd, " -spmap %S ", spmap);
if(MAJSTRGETLEN(cons))
ajFmtPrintAppS(&cmd, "-cons %S", cons);
if(maxsize != -1)
ajFmtPrintAppS(&cmd, " -maxsize %d ", maxsize);
if(p > 0)
ajFmtPrintAppS(&cmd, " -p %d ", p);
if(time > 0)
ajFmtPrintAppS(&cmd, " -time %d ", time);
if(MAJSTRGETLEN(sf))
ajFmtPrintAppS(&cmd, " -sf %S", sf);
if(heapsize != 64)
ajFmtPrintAppS(&cmd, " -heapsize %d ", heapsize);
if(xbranch)
ajFmtPrintAppS(&cmd, " -x_branch");
if(wbranch)
ajFmtPrintAppS(&cmd, " -w_branch");
if(bfactor != 3)
ajFmtPrintAppS(&cmd, " -bfactor %d ", bfactor);
if(ajSeqsetIsProt(dataset))
ajFmtPrintAppS(&cmd, "-protein ");
else
ajFmtPrintAppS(&cmd, "-dna ");
ajFmtPrintAppS(&cmd, " -text");
ajFmtPrintAppS(&cmd, " > %S ", ajFileGetNameS(outtext));
/* 4. Close files from ACD before calling meme */
ajFileClose(&bfile);
ajFileClose(&plib);
/* 5. Call meme */
/* ajFmtPrint("\n%S\n", cmd); */
system(ajStrGetPtr(cmd));
/* 6. Exit cleanly */
ajSeqsetDel(&dataset);
ajStrDel(&cons);
ajStrDel(&sf);
ajStrDel(&mod);
ajStrDel(&prior);
ajStrDel(&spmap);
ajStrDel(&cmd);
ajStrDel(&ssname);
ajStrDel(&tmp);
ajFileClose(&bfile);
ajFileClose(&plib);
ajFileClose(&outtext);
ajSeqoutDel(&outseq);
embExit();
return 0;
}
int main(int argc, char **argv)
{
/* ACD data item variables */
AjPFile hmmfile = NULL;
AjPSeqset seqfile = NULL;
AjPFile mapali = NULL;
AjPFile withali = NULL;
AjPAlign o = NULL;
AjBool m = ajFalse;
AjBool q = ajFalse;
/* Housekeeping variables */
AjPStr cmd = NULL;
AjPStr fmt = NULL;
AjBool fmtok = ajFalse;
AjPStr rnd = NULL;
AjPSeqout rndo = NULL;
/* ACD file processing */
embInitPV("ehmmalign",argc,argv,"HMMERNEW",VERSION);
hmmfile = ajAcdGetInfile("hmmfile");
seqfile = ajAcdGetSeqset("seqfile");
mapali = ajAcdGetInfile("mapali");
withali = ajAcdGetInfile("withali");
o = ajAcdGetAlign("o");
m = ajAcdGetBoolean("m");
q = ajAcdGetBoolean("q");
/* MAIN APPLICATION CODE */
/* 1. Housekeeping */
cmd = ajStrNew();
fmt = ajStrNew();
rnd = ajStrNew();
/* 2. Re-write seqfile to a temporary file in a format (fasta) HMMER can understand.
We cannot just pass the name of seqfile to HMMER as the name provided might be a
USA which HMMER would not understand. */
ajFilenameSetTempname(&rnd);
rndo = ajSeqoutNew();
if(!ajSeqoutOpenFilename(rndo, rnd))
ajFatal("Terminal ajSeqFileNewOut failure. Email EMBOSS helpdesk!\n");
ajSeqoutSetFormatC(rndo, "fasta");
ajSeqoutWriteSet(rndo, seqfile);
ajSeqoutClose(rndo);
ajSeqoutDel(&rndo);
/* 3. Build hmmalign command line */
/* Command line is built in this order:
i. Application name.
ii. HMMER 'options' (in order they appear in ACD file)
iii.HMMER 'options' (that don't appear in ACD file)
iv. HMMER & new parameters.
*/
ajFmtPrintS(&cmd, "%S ", ajAcdGetpathC("hmmalign"));
if(mapali)
ajFmtPrintAppS(&cmd, " --mapali %s ", ajFileGetNameC(mapali));
if(withali)
ajFmtPrintAppS(&cmd, " --withali %s ", ajFileGetNameC(withali));
if(m)
ajStrAppendC(&cmd, " -m ");
if(q)
ajStrAppendC(&cmd, " -q ");
/* Ensure output alignment is in user-specified format. */
fmtok=ajTrue;
ajStrAssignS(&fmt, ajAlignGetFormat(o));
/* fasta and a2m are identical formats. */
if(ajStrMatchC(fmt, "fasta"))
ajStrAssignC(&fmt, "A2M");
else if(ajStrMatchC(fmt, "a2m"))
ajStrAssignC(&fmt, "A2M");
else if(ajStrMatchC(fmt, "msf"))
ajStrAssignC(&fmt, "MSF");
else if(ajStrMatchC(fmt, "phylip"))
ajStrAssignC(&fmt, "PHYLIP");
/* hmmer also supports stockholm, SELEX & Clustal output, EMBOSS does not.
EMBOSS supports unknown/multiple/simple and srs output, hmmer does not. */
else
fmtok = ajFalse;
if(!fmtok)
{
/* This could be replaced with code to reformat the file. */
ajWarn("Specified output alignment format ('o' ACD option) is "
"not understood by HMMER. Using stockholm format instead.");
ajStrAssignC(&fmt, "Stockholm");
}
/* rnd is the name of the rewritten seqfile. MUST specify FASTA format explicitly. */
ajFmtPrintAppS(&cmd, " --informat FASTA --outformat %S -o %s %s %S",
fmt,
ajAlignGetFilename(o),
ajFileGetNameC(hmmfile),
rnd);
/* 4. Close ACD files */
ajFileClose(&hmmfile);
ajSeqsetDel(&seqfile);
ajFileClose(&mapali);
ajFileClose(&withali);
ajAlignClose(o);
ajAlignDel(&o);
/* 5. Call hmmalign */
ajFmtPrint("\n%S\n\n", cmd);
ajSysExecS(cmd);
/* 6. Exit cleanly */
ajSysFileUnlinkS(rnd);
ajStrDel(&cmd);
ajStrDel(&fmt);
ajStrDel(&rnd);
embExit();
return 0;
}
int main(int argc, char **argv)
{
AjPSeqset seqset = NULL;
AjPStr cl = NULL;
AjPSeqout seqout = NULL;
AjBool pc = ajFalse;
AjBool sigp = ajFalse;
AjBool plot = ajFalse;
const AjPStr ofn = NULL;
AjPStr fn = NULL;
AjPStr stmp = NULL;
AjPFile outf = NULL;
embInitPV("eprop", argc, argv, "CBSTOOLS", VERSION);
seqset = ajAcdGetSeqset("sequence");
outf = ajAcdGetOutfile("outfile");
plot = ajAcdGetBoolean("plot");
pc = ajAcdGetBoolean("pcprediction");
sigp = ajAcdGetBoolean("signalp");
cl = ajStrNewC("prop ");
fn = ajStrNew();
stmp = ajStrNew();
ajFilenameSetTempname(&fn);
seqout = ajSeqoutNew();
if(!ajSeqoutOpenFilename(seqout, fn))
ajFatal("Cannot open temporary file %S",fn);
ajSeqoutSetFormatC(seqout, "fasta");
ajSeqoutWriteSet(seqout,seqset);
ajSeqoutClose(seqout);
if(plot)
ajStrAppendC(&cl," -g");
if(pc)
ajStrAppendC(&cl," -p");
if(sigp)
ajStrAppendC(&cl," -s");
ajFmtPrintS(&stmp," %S",fn);
ajStrAppendS(&cl,stmp);
ofn = ajFileGetNameS(outf);
ajFmtPrintS(&stmp," > %S",ofn);
ajStrAppendS(&cl,stmp);
ajFileClose(&outf);
#if 0
ajFmtPrint("%S\n",cl);
#endif
#if 1
if(system(ajStrGetPtr(cl)) == -1)
ajFatal("Command %S failed",cl);
#endif
ajSysFileUnlink(fn);
ajStrDel(&cl);
ajStrDel(&fn);
ajStrDel(&stmp);
ajSeqoutDel(&seqout);
ajSeqsetDel(&seqset);
embExit();
return 0;
}
int main(int argc, char **argv)
{
AjPStr cl = NULL;
AjPSeqset seqset = NULL;
AjPSeqout seqout = NULL;
AjPStr stmp = NULL;
AjPStr email = NULL;
AjPStr fmt = NULL;
AjPStr trtab = NULL;
const AjPStr ofn = NULL;
AjPStr *applist = NULL;
AjBool crc = ajFalse;
AjBool alt = ajFalse;
AjBool iprlook = ajFalse;
AjBool goterms = ajFalse;
ajint trlen = 0;
ajuint i = 0;
ajuint lcnt = 0;
AjPStr fn = NULL;
AjPFile outf = NULL;
embInitPV("eiprscan", argc, argv, "IPRSCAN", VERSION);
seqset = ajAcdGetSeqset("sequence");
email = ajAcdGetString("email");
crc = ajAcdGetBoolean("crc");
applist = ajAcdGetList("appl");
fmt = ajAcdGetListSingle("format");
trtab = ajAcdGetListSingle("trtable");
trlen = ajAcdGetInt("trlen");
alt = ajAcdGetBoolean("altjobs");
iprlook = ajAcdGetBoolean("iprlookup");
goterms = ajAcdGetBoolean("goterms");
outf = ajAcdGetOutfile("outfile");
stmp = ajStrNew();
cl = ajStrNewC("iprscan -cli");
fn = ajStrNew();
ajFilenameSetTempname(&fn);
seqout = ajSeqoutNew();
if(!ajSeqoutOpenFilename(seqout, fn))
ajFatal("Cannot open temporary file %S",fn);
ajSeqoutSetFormatC(seqout, "fasta");
ajSeqoutWriteSet(seqout,seqset);
ajSeqoutClose(seqout);
ajFmtPrintS(&stmp," -i %S",fn);
ajStrAppendS(&cl,stmp);
if(!ajSeqsetIsProt(seqset))
ajStrAppendC(&cl," -seqtype n");
if(!crc)
ajStrAppendC(&cl," -nocrc");
if(ajStrGetLen(email))
{
ajFmtPrintS(&stmp," -email %S",email);
ajStrAppendS(&cl,stmp);
}
i = 0;
lcnt = 0;
while(applist[i])
{
++lcnt;
++i;
}
i = 0;
while(applist[i])
{
if(ajStrMatchC(applist[i],"all"))
{
if(lcnt != 1)
ajFatal("Cannot specify 'all' with multiple "
"applications");
++i;
continue;
}
ajFmtPrintS(&stmp," -appl %S",applist[i]);
ajStrAppendS(&cl,stmp);
++i;
}
ajFmtPrintS(&stmp," -format %S",fmt);
ajStrAppendS(&cl,stmp);
ajFmtPrintS(&stmp," -trtable %S",trtab);
ajStrAppendS(&cl,stmp);
ajFmtPrintS(&stmp," -trlen %d",trlen);
ajStrAppendS(&cl,stmp);
if(alt)
ajStrAppendC(&cl," -altjobs");
if(iprlook)
ajStrAppendC(&cl," -iprlookup");
if(goterms)
ajStrAppendC(&cl," -goterms");
ofn = ajFileGetNameS(outf);
ajFmtPrintS(&stmp," -o %S",ofn);
ajFileClose(&outf);
ajStrAppendS(&cl,stmp);
#if 0
ajFmtPrint("%S\n",cl);
#endif
#if 1
system(ajStrGetPtr(cl));
#endif
ajSysFileUnlink(fn);
ajStrDel(&cl);
ajStrDel(&fn);
ajStrDel(&stmp);
ajStrDel(&email);
ajStrDel(&fmt);
ajStrDel(&trtab);
ajStrDelarray(&applist);
ajSeqoutDel(&seqout);
ajSeqsetDel(&seqset);
embExit();
return 0;
}
int main(int argc, char **argv)
{
AjPSeqset seqset = NULL;
AjPStr cl = NULL;
AjPSeqout seqout = NULL;
float thresh = 0.;
AjBool netphos = ajFalse;
AjPStr format = NULL;
AjBool plot = ajFalse;
AjPStr fn = NULL;
AjPStr stmp = NULL;
AjPStr outfname = NULL;
embInitPV("eyinoyang", argc, argv, "CBSTOOLS",VERSION);
seqset = ajAcdGetSeqset("sequence");
outfname= ajAcdGetOutfileName("outfile");
plot = ajAcdGetBoolean("plot");
netphos = ajAcdGetBoolean("netphos");
thresh = ajAcdGetFloat("threshold");
format = ajAcdGetListSingle("format");
cl = ajStrNewS(ajAcdGetpathC("yinOyang"));
fn = ajStrNew();
stmp = ajStrNew();
ajFilenameSetTempname(&fn);
seqout = ajSeqoutNew();
if(!ajSeqoutOpenFilename(seqout, fn))
ajFatal("Cannot open temporary file %S",fn);
ajSeqoutSetFormatC(seqout, "fasta");
ajSeqoutWriteSet(seqout,seqset);
ajSeqoutClose(seqout);
if(ajStrMatchC(format,"short"))
ajStrAppendC(&cl," -f s");
else if(ajStrMatchC(format,"long"))
ajStrAppendC(&cl," -f l");
if(plot)
ajStrAppendC(&cl," -g");
if(netphos)
ajStrAppendC(&cl," -y");
if(netphos)
{
ajFmtPrintS(&stmp," -t %f",thresh);
ajStrAppendS(&cl,stmp);
}
ajFmtPrintS(&stmp," %S",fn);
ajStrAppendS(&cl,stmp);
#if 0
ajFmtPrint("%`S\n",cl);
#endif
#if 1
ajSysExecOutnameAppendS(cl, outfname);
#endif
ajSysFileUnlinkS(fn);
ajStrDel(&cl);
ajStrDel(&fn);
ajStrDel(&stmp);
ajStrDel(&format);
ajSeqoutDel(&seqout);
ajSeqsetDel(&seqset);
ajStrDel(&outfname);
embExit();
return 0;
}
