/* @funcstatic seqwords_TermsRead *********************************************
**
** Read the next Terms object from a file in embl-like format. The search
** terms are modified with a leading and trailing space.
**
** @param [r] inf [AjPFile] Input file stream
** @param [w] thys [AjPTerms*] Terms object
**
** @return [AjBool] True on succcess
** @@
*****************************************************************************/
static AjBool seqwords_TermsRead(AjPFile inf,
AjPTerms *thys)
{
AjPStr line =NULL; /* Line of text. */
AjPStr temp =NULL;
AjPList list_terms =NULL; /* List of keywords for a scop node*/
AjBool ok =ajFalse;
AjPStr type = NULL;
/* Memory management */
(*thys)=seqwords_TermsNew();
list_terms = ajListstrNew();
line = ajStrNew();
type = ajStrNew();
/* Read first line. */
ok = ajReadlineTrim(inf,&line);
while(ok && !ajStrPrefixC(line,"//"))
{
if(ajStrPrefixC(line,"XX"))
{
ok = ajReadlineTrim(inf,&line);
continue;
}
else if(ajStrPrefixC(line,"TY"))
{
ajFmtScanS(line, "%*s %S", &type);
if(ajStrMatchC(type, "SCOP"))
(*thys)->Type = ajSCOP;
else if(ajStrMatchC(type, "CATH"))
(*thys)->Type = ajCATH;
}
else if(ajStrPrefixC(line,"CL"))
{
ajStrAssignC(&(*thys)->Class,ajStrGetPtr(line)+3);
ajStrRemoveWhiteExcess(&(*thys)->Class);
}
else if(ajStrPrefixC(line,"AR"))
{
ajStrAssignC(&(*thys)->Architecture,ajStrGetPtr(line)+3);
ajStrRemoveWhiteExcess(&(*thys)->Architecture);
}
else if(ajStrPrefixC(line,"TP"))
{
ajStrAssignC(&(*thys)->Topology,ajStrGetPtr(line)+3);
ajStrRemoveWhiteExcess(&(*thys)->Topology);
}
else if(ajStrPrefixC(line,"FO"))
{
ajStrAssignC(&(*thys)->Fold,ajStrGetPtr(line)+3);
while(ajReadlineTrim(inf,&line))
{
if(ajStrPrefixC(line,"XX"))
break;
ajStrAppendC(&(*thys)->Fold,ajStrGetPtr(line)+3);
}
ajStrRemoveWhiteExcess(&(*thys)->Fold);
}
else if(ajStrPrefixC(line,"SF"))
{
ajStrAssignC(&(*thys)->Superfamily,ajStrGetPtr(line)+3);
while(ajReadlineTrim(inf,&line))
{
if(ajStrPrefixC(line,"XX"))
break;
ajStrAppendC(&(*thys)->Superfamily,ajStrGetPtr(line)+3);
}
ajStrRemoveWhiteExcess(&(*thys)->Superfamily);
}
else if(ajStrPrefixC(line,"FA"))
{
ajStrAssignC(&(*thys)->Family,ajStrGetPtr(line)+3);
while(ajReadlineTrim(inf,&line))
{
if(ajStrPrefixC(line,"XX"))
break;
ajStrAppendC(&(*thys)->Family,ajStrGetPtr(line)+3);
}
ajStrRemoveWhiteExcess(&(*thys)->Family);
}
else if(ajStrPrefixC(line,"TE"))
{
/* Copy and clean up term. */
temp = ajStrNew();
ajStrAssignC(&temp,ajStrGetPtr(line)+3);
ajStrRemoveWhiteExcess(&temp);
/* Append a leading and trailing space to search term*/
ajStrAppendK(&temp, ' ');
ajStrInsertC(&temp, 0, " ");
/* Add the current term to the list. */
ajListstrPush(list_terms,temp);
}
ok = ajReadlineTrim(inf,&line);
}
if(!ok)
{
/* Clean up. */
ajListstrFree(&list_terms);
ajStrDel(&line);
/* Return. */
return ajFalse;
}
/* Convert the AjPList of terms to array of AjPSeq's. */
if(!((*thys)->N=ajListstrToarray((AjPList)list_terms,&(*thys)->Keywords)))
ajWarn("Zero sized list of terms passed into seqwords_TermsRead");
/* Clean up. Free the list (not the nodes!). */
ajListstrFree(&list_terms);
ajStrDel(&line);
ajStrDel(&type);
return ajTrue;
}
static void acdrelations_readdatfile
(AjPFile inf,
PEdam *P)
{
AjPStr line = NULL;
const AjPStr tok = NULL;
const AjPStr subtok = NULL;
AjPStr strtmp = NULL;
AjPList strlist = NULL;
AjPStr acdtype = NULL;
AjPStr relations = NULL;
PEdamdat dattmp = NULL;
AjPList datlist = NULL;
if(!P)
ajFatal("Null arg error 1 in acdrelations_readdatfile");
if(!inf)
ajFatal("Null arg error 3 in acdrelations_readdatfile");
/* Allocate memory */
line = ajStrNew();
acdtype = ajStrNew();
relations = ajStrNew();
datlist = ajListNew();
/* Read data from file */
while(ajReadline(inf,&line))
{
/* Discard comment lines */
if(ajStrPrefixC(line,"#"))
continue;
/* Tokenise line, delimited by '|'.
Parse first token (ACD datatype ) */
ajStrAssignS(&acdtype, ajStrParseC(line, "|"));
/* Parse second token (EDAM relations: value ) */
ajStrAssignS(&relations, ajStrParseC(NULL, "|"));
/* Parse third token (attribute:value strings block) */
tok = ajStrParseC(NULL, "|");
/* Create new string list */
strlist = ajListstrNew();
/* Tokenise third token itself into tokens delimited by ' ' (space)
Parse tokens (individual attribute:value strings)*/
if((subtok=ajStrParseC(tok, ";")))
{
strtmp = ajStrNew();
ajStrAssignS(&strtmp, subtok);
ajStrRemoveWhite(&strtmp);
ajListstrPushAppend(strlist, strtmp);
while((subtok=ajStrParseC(NULL, ";")))
{
strtmp = ajStrNew();
ajStrAssignS(&strtmp, subtok);
ajStrRemoveWhite(&strtmp);
ajListstrPushAppend(strlist, strtmp);
}
}
/* Write PEdamdat structure & push onto list */
dattmp = ajEdamdatNew();
ajStrRemoveWhite(&acdtype);
ajStrAssignS(&dattmp->acdtype, acdtype);
ajStrAssignS(&dattmp->edam, relations);
dattmp->n = ajListstrToarray(strlist, &dattmp->acdattr);
ajListPushAppend(datlist, dattmp);
/* Clear nodes (but not strings) from string list */
ajListstrFree(&strlist);
}
/* Write PEdam structure */
((*P)->n) = ajListToarray(datlist, (void***) &((*P)->dat));
/* Free memory */
ajStrDel(&line);
ajStrDel(&acdtype);
ajStrDel(&relations);
ajListFree(&datlist);
return;
}
/* @funcstatic acdrelations_procacdfile ***************************************
**
** Process ACD file and write new ACD file with new relations: attributes
** added (replaced if necessary).
**
** @param [r] inf [AjPFile] ACD input file
** @param [r] outf [AjPFile] ACD output file
** @param [r] P [PEdam] edam object
** @param [r] T [PKtype] ktype object
** @return [void]
** @@
******************************************************************************/
static void acdrelations_procacdfile
(AjPFile inf,
AjPFile outf,
PEdam P,
PKtype T)
{
AjPStr line = NULL;
AjPStr tok = NULL;
AjPStr acdtype = NULL;
AjPStr strtmp = NULL;
AjPList strlist = NULL;
AjPStr *strarr = NULL;
ajint nstr = 0;
/* Allocate memory */
line = ajStrNew();
tok = ajStrNew();
acdtype = ajStrNew();
strlist = ajListstrNew();
/* Read next line */
while(ajReadline(inf,&line))
{
ajFmtScanS(line, "%S", &tok);
/* Write application definition or section definition out as-is */
if(ajStrMatchC(tok, "application:") ||
ajStrMatchC(tok, "section:"))
{
ajFmtPrintF(outf, "%S", line);
while(ajReadline(inf,&line))
{
ajFmtPrintF(outf, "%S", line);
ajFmtScanS(line, "%S", &tok);
if(ajStrMatchC(tok, "]"))
break;
}
}
/* Write variables, endsection definitions and comments out as-is */
else if(ajStrMatchC(tok, "variable:") ||
ajStrMatchC(tok, "endsection:") ||
ajStrMatchC(tok, "#"))
ajFmtPrintF(outf, "%S", line);
/* Write out blank lines as-is */
else if (!ajFmtScanS(line, "%S", &tok))
ajFmtPrintF(outf, "%S", line);
/* Process data definition */
else
/* First line of data definition */
{
/* Process and write datatype line */
ajFmtPrintF(outf, "%S", line);
ajFmtScanS(line, "%S", &acdtype);
ajStrRemoveSetC(&acdtype, ":");
/* Process subsequent (attribute) lines */
while(ajReadline(inf,&line))
{
strtmp = ajStrNew();
ajStrAssignS(&strtmp, line);
ajStrRemoveWhite(&strtmp);
ajListstrPushAppend(strlist, strtmp);
ajFmtScanS(line, "%S", &tok);
/* Reached end of data definition */
if(ajStrMatchC(tok, "]"))
{
nstr = ajListstrToarray(strlist, &strarr);
/* Write relations: line */
acdrelations_writerelations(outf, acdtype, strarr, nstr, P, T);
AJFREE(strarr);
ajListstrFreeData(&strlist);
strlist = ajListstrNew();
ajFmtPrintF(outf, "%S", line);
break;
}
/* Ignore existing relations: lines */
else if(ajStrMatchC(tok, "relations:"))
continue;
ajFmtPrintF(outf, "%S", line);
}
}
}
/* Free memory */
ajStrDel(&line);
ajStrDel(&tok);
ajStrDel(&acdtype);
ajListstrFreeData(&strlist);
return;
}
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;
}
