int main(ajint argc, char **argv)
{
AjPList ccfin = NULL; /* List of CCF (input) files. */
AjPDir pdbin = NULL; /* Path of pdb input files. */
AjPStr pdbprefix = NULL; /* Prefix of pdb input files. */
AjPStr pdb_name = NULL; /* Full name (path/name/extension) of
pdb format input file. */
AjPDirout ccfout = NULL; /* Path of coordinate output file. */
AjPStr randomname = NULL; /* Name for temp file tempf. */
AjPStr ccf_this = NULL;
AjPStr exec = NULL;
AjPStr naccess_str = NULL;
AjPStr line = NULL;
AjPStr syscmd = NULL; /* Command line arguments. */
AjPStr *mode = NULL; /* Mode of operation from acd. */
AjPFile errf = NULL; /* pdbplus error file pointer. */
AjPFile serrf = NULL; /* stride error file pointer. */
AjPFile nerrf = NULL; /* stride error file pointer. */
AjPFile tempf = NULL; /* Temp file for holding STRIDE output. */
AjPFile ccf_inf = NULL; /* Protein coordinate input file. */
AjPFile ccf_outf = NULL; /* Protein coordinate output file. */
AjIList iter = NULL;
AjBool done_naccess= ajFalse;
AjBool done_stride = ajFalse;
AjBool found = ajFalse;
AjPResidue temp_res = NULL; /* Pointer to Residue object. */
AjPPdb pdb_old = NULL; /* Pointer to PDB object - without new
stride elements. */
AjPPdb pdb = NULL; /* Pointer to PDB object. */
ajint idn = 0; /* Chain identifier as a number (1,2,...) */
ajint chain_num = 0; /* Chain identifier index (0,1,...). */
ajint tS = 0; /* User-defined threshold size for SSEs. */
ajint nostride = 0; /* No. times stride failed */
ajint nonaccess = 0; /* No. times naccess failed */
ajint nofile = 0; /* No. times of file error */
/* Variables for each item that will be parsed from the ASG line. */
AjPStr res = NULL; /* Residue id from STRIDE ASG line (ALA etc). */
AjPStr res_num = NULL; /* PDB residue number from STRIDE ASG line. */
char pcid = ' '; /* Protein chain identifier from STRIDE or
NACESS output (A,B, etc). */
char ss = ' '; /* One-letter secondary structure code from
STRIDE ASG line. */
float ph = 0.0; /* Phi angle from STRIDE ASG line. */
float ps = 0.0; /* Psi angle from STRIDE ASG line. */
float sa = 0.0; /* Residue solvent accessible area from STRIDE
ASG line. */
float f1 = 0;
float f2 = 0;
float f3 = 0;
float f4 = 0;
float f5 = 0;
float f6 = 0;
float f7 = 0;
float f8 = 0;
float f9 = 0;
float f10 = 0;
/* Allocate strings; this section is used for variables that are
allocated once only. */
pdb_name = ajStrNew();
res = ajStrNew();
res_num = ajStrNew();
randomname = ajStrNew();
syscmd = ajStrNew();
line = ajStrNew();
naccess_str = ajStrNew();
exec = ajStrNew();
/* Read data from acd. */
embInitPV("pdbplus",argc,argv,"STRUCTURE",VERSION);
ccfin = ajAcdGetDirlist("ccfinpath");
pdbin = ajAcdGetDirectory("pdbindir");
pdbprefix = ajAcdGetString("pdbprefix");
ccfout = ajAcdGetOutdir("ccfoutdir");
mode = ajAcdGetList("mode");
errf = ajAcdGetOutfile("logfile");
if(ajStrGetCharFirst(*mode) != '2')
serrf = ajAcdGetOutfile("slogfile");
if(ajStrGetCharFirst(*mode) != '1')
nerrf = ajAcdGetOutfile("nlogfile");
tS = ajAcdGetInt("thresholdsize");
ajRandomSeed();
ajFilenameSetTempname(&randomname);
/*
** Start of main application loop.
** Process each PDB/ protein coordinate file (EMBL format) in turn.
*/
while(ajListPop(ccfin,(void **)&ccf_this))
{
/* Open protein coordinate file. If it cannot be opened, write a
message to the error file, delete ccf_this and continue. */
if((ccf_inf = ajFileNewInNameS(ccf_this)) == NULL)
{
ajWarn("%s%S\n//\n",
"clean coordinate file not found: ", ccf_this);
ajFmtPrintF(errf, "%s%S\n//\n",
"clean coordinate file not found: ", ccf_this);
ajStrDel(&ccf_this);
nofile++;
continue;
}
ajFmtPrint("Processing %S\n", ccf_this);
fflush(stdout);
/* Parse protein coordinate data (from clean format file) into
AjPPdb object. ajPdbReadAllModelsNew will create the AjPPdb object. */
if(!(pdb_old=ajPdbReadAllModelsNew(ccf_inf)))
{
ajWarn("ERROR Clean coordinate file read"
"error: %S\n//\n", ccf_this);
ajFmtPrintF(errf, "ERROR Clean coordinate file read"
"error: %S\n//\n", ccf_this);
ajFileClose(&ccf_inf);
ajStrDel(&ccf_this);
nofile++;
continue;
}
ajFileClose(&ccf_inf);
ajPdbCopy(&pdb, pdb_old);
ajPdbDel(&pdb_old);
/* Construct name of corresponding PDB file.
NACCESS does *not* generate an output file if the path is './' e.g.
naccess ./1rbp.ent , therefore replace './' with null. */
ajStrAssignS(&pdb_name, ajDirGetPath(pdbin));
if(ajStrMatchC(pdb_name, "./") || ajStrMatchC(pdb_name, "."))
ajStrAssignC(&pdb_name, "");
ajStrAppendS(&pdb_name, pdbprefix);
ajStrFmtLower(&pdb->Pdb);
ajStrAppendS(&pdb_name, pdb->Pdb);
ajStrAppendC(&pdb_name, ".");
ajStrAppendS(&pdb_name, ajDirGetExt(pdbin));
/* Check corresponding PDB file exists for reading using ajFileStat. */
if(!(ajFilenameExistsRead(pdb_name)))
{
ajFmtPrintF(errf, "%s%S\n//\n", "PDB file not found: ", pdb_name);
ajWarn("%s%S\n//\n", "PDB file not found: ", pdb_name);
ajStrDel(&ccf_this);
ajPdbDel(&pdb);
nofile++;
continue;
}
if(ajStrGetCharFirst(*mode) != '2')
{
/*
** Create a string containing the STRIDE command line (it needs
** PDB file name & name of temp output file).
** Call STRIDE by using ajSystem.
*/
ajFmtPrintS(&syscmd, "%S %S -f%S >> %s 2>&1",
ajAcdGetpathC("stride"),
pdb_name, randomname, ajFileGetNameC(serrf));
ajFmtPrint("%S %S -f%S >> %s 2>&1\n",
ajAcdGetpathC("stride"),
pdb_name, randomname,ajFileGetNameC(serrf));
system(ajStrGetPtr(syscmd));
/* Open the stride output file */
if (((tempf = ajFileNewInNameS(randomname)) == NULL))
{
ajWarn("%s%S\n//\n",
"no stride output for: ", pdb_name);
ajFmtPrintF(errf, "%s%S\n//\n",
"no stride output for: ", pdb_name);
nostride++;
ajStrDel(&ccf_this);
ajPdbDel(&pdb);
continue;
}
else
ajFmtPrintF(errf, "%s%S\n//\n",
"stride output for: ", pdb_name);
done_stride = ajFalse;
/* Parse STRIDE output from temp output file a line at a time. */
while(ajReadlineTrim(tempf,&line))
{
if(ajStrPrefixC(line,"ASG"))
{
ajFmtScanS(line, "%*S %S %c %S %*d %c %*S %f %f %f %*S",
&res, &pcid, &res_num, &ss, &ph, &ps, &sa);
/*
** Populate pdbplus object with the data from this parsed
** line. This means first identifying the chain, then
** finding the residue.
*/
/* Determine the chain number. ajDmxPdbplusChain does not
recognise '-', so change '-' to '.' */
if (pcid == '-')
pcid = '.';
/* Get chain number from the chain identifier. */
if(!ajPdbChnidToNum(pcid, pdb, &idn))
{
ajWarn("Could not convert chain id %c to chain"
" number in pdb file %S\n//\n",
pcid, pdb_name);
ajFmtPrintF(errf, "Could not convert chain id %c "
"to chain number in pdb file %S\n//\n",
pcid, pdb_name);
continue;
}
/*
** The chain number that will get written starts at 1, but
** we want an index into an array which must start at 0,
** so subtract 1 from the chain number to get the index.
*/
chain_num = idn-1;
/*
** Iiterate through the list of residues in the Pdb object,
** found switches to true when first residue corresponding
** to the line is found.
*/
/* iter = ajListIterNewread(pdb->Chains[chain_num]->Atoms); */
iter = ajListIterNewread(pdb->Chains[chain_num]->Residues);
found = ajFalse;
while((temp_res = (AjPResidue)ajListIterGet(iter)))
{
/* If we have found the residue we want */
if((ajStrMatchS(res_num, temp_res->Pdb) &&
ajStrMatchS(res, temp_res->Id3)))
{
done_stride = ajTrue;
found = ajTrue;
temp_res->eStrideType = ss;
temp_res->Phi = ph;
temp_res->Psi = ps;
temp_res->Area = sa;
}
/* If the matching residue has been processed
move on to next ASG line, next residue. */
else if(found == ajTrue)
break;
else
/* Matching residue not found yet. */
continue;
}
ajListIterDel(&iter);
} /* End of if ASG loop. */
} /* End of while line loop. */
if(done_stride)
ajFmtPrintF(errf, "%s%S\n//\n",
"stride data for: ", pdb_name);
else
{
ajFmtPrintF(errf, "%s%S\n//\n",
"no stride data for: ", pdb_name);
ajWarn("%s%S\n//\n", "no stride data for: ", pdb_name);
nostride++;
}
/* Close STRIDE temp file. & tidy up. */
ajFileClose(&tempf);
/* Remove temporary file (stride output file). */
ajFmtPrintS(&exec, "rm %S", randomname);
ajSysSystem(exec);
/*
** Calculate element serial numbers (eStrideNum)& ammend residue
** objects, count no's of elements and ammend chain object
** (numHelices, num Strands).
*/
pdbplus_sort(pdb, tS);
}
if(ajStrGetCharFirst(*mode) != '1')
{
/*
** Create a string containing the NACCESS command line (it needs
** PDB file name & name of temp output file) & call NACCESS.
** If e.g. /data/structure/pdbfred.ent was parsed and the program
** was run from /stuff, then /stuff/fred.asa and /stuff/fred.rsa
** would be written. These must be deleted once parsed (only
** use the .rsa file here).
*/
ajFmtPrintS(&syscmd, "%S %S >> %s 2>&1",
ajAcdGetpathC("naccess"), pdb_name,
ajFileGetNameC(nerrf));
ajFmtPrint("%S %S >> %s 2>&1\n",
ajAcdGetpathC("naccess"), pdb_name,
ajFileGetNameC(nerrf));
system(ajStrGetPtr(syscmd));
ajStrAssignS(&naccess_str, pdbprefix);
ajStrAppendS(&naccess_str, pdb->Pdb);
ajStrAppendC(&naccess_str, ".rsa");
/* Open the NACCESS output file. */
if (((tempf = ajFileNewInNameS(naccess_str)) == NULL))
{
ajFmtPrintF(errf, "%s%S\n//\n",
"no naccess output for: ", pdb_name);
ajWarn("%s%S\n//\n", "no naccess output for: ", pdb_name);
nonaccess++;
ajStrDel(&ccf_this);
ajPdbDel(&pdb);
continue;
}
else
ajFmtPrintF(errf, "%s%S\n//\n",
"naccess output for: ", pdb_name);
done_naccess = ajFalse;
/* Parse NACCESS output from temp output file a line at a time. */
while(ajReadlineTrim(tempf,&line))
{
if(ajStrPrefixC(line,"RES"))
{
/* Read data from lines. */
if((pcid = line->Ptr[8]) == ' ')
ajFmtScanS(line, "%*S %S %S %f %f %f "
"%f %f %f %f %f %f %f",
&res, &res_num, &f1, &f2, &f3, &f4, &f5,
&f6, &f7, &f8, &f9, &f10);
else
ajFmtScanS(line, "%*S %S %*c %S %f %f "
"%f %f %f %f %f %f %f %f",
&res, &res_num, &f1, &f2, &f3, &f4, &f5,
&f6, &f7, &f8, &f9, &f10);
/* Identify the chain, then finding all the residues
corresponding to the residue. */
/* Get the chain number from the chain identifier. */
if(!ajPdbChnidToNum(pcid, pdb, &idn))
{
ajWarn("Could not convert chain id %c to chain"
" number in pdb file %S\n//\n",
pcid, pdb_name);
ajFmtPrintF(errf, "Could not convert chain id"
" %c to chain number in pdb file %S\n//\n",
pcid, pdb_name);
continue;
}
/*
** Chain number will start at 1, but we want an index
** into an array which must start at 0, so subtract 1
** from the chain number to get the index.
*/
chain_num = idn-1;
/*
** Iiterate through the list of residues in the Pdb object,
** temp_res is an AjPResidue used to point to the current
** residue.
** ajBool found switches to true when first residue
** corresponding to the line is found.
*/
iter = ajListIterNewread(pdb->Chains[chain_num]->Residues);
found = ajFalse;
while((temp_res = (AjPResidue)ajListIterGet(iter)))
{
/* If we have found the residue we want, write the residue
object. */
if((ajStrMatchS(res_num, temp_res->Pdb) &&
ajStrMatchS(res, temp_res->Id3)))
{
found = ajTrue;
done_naccess = ajTrue;
temp_res->all_abs = f1;
temp_res->all_rel = f2;
temp_res->side_abs = f3;
temp_res->side_rel = f4;
temp_res->main_abs = f5;
temp_res->main_rel = f6;
temp_res->npol_abs = f7;
temp_res->npol_rel = f8;
temp_res->pol_abs = f9;
temp_res->pol_rel = f10;
}
/* If the matching residues have all been processed.
move on to next ASG line, next residue. */
else if(found == ajTrue)
break;
else
/* Matching residues not found yet, move on to next
residue. */
continue;
}
ajListIterDel(&iter);
}
}
if(done_naccess)
ajFmtPrintF(errf, "%s%S\n//\n",
"naccess data for: ", pdb_name);
else
{
ajFmtPrintF(errf, "%s%S\n//\n",
"no naccess data for: ", pdb_name);
ajWarn("%s%S\n//\n", "no naccess data for: ", pdb_name);
nonaccess++;
}
/* Remove temporary file (naccess output files). */
ajFileClose(&tempf);
ajFmtPrintS(&exec, "rm %S", naccess_str);
ajSysSystem(exec);
ajStrAssignS(&naccess_str, pdbprefix);
ajStrAppendS(&naccess_str, pdb->Pdb);
ajStrAppendC(&naccess_str, ".asa");
ajFmtPrintS(&exec, "rm %S", naccess_str);
ajSysSystem(exec);
ajStrAssignS(&naccess_str, pdbprefix);
ajStrAppendS(&naccess_str, pdb->Pdb);
ajStrAppendC(&naccess_str, ".log");
ajFmtPrintS(&exec, "rm %S", naccess_str);
ajSysSystem(exec);
}
/* Open CCF (output) file. */
ccf_outf = ajFileNewOutNameDirS(pdb->Pdb, ccfout);
/* Write AjPPdb object to the output file in clean format. */
if(!ajPdbWriteAll(ccf_outf, pdb))
{
ajWarn("%s%S\n//\n","Could not write results file for: ",
pdb->Pdb);
ajFmtPrintF(errf,"%s%S\n//\n","Could not write results file for ",
pdb->Pdb);
}
ajFileClose(&ccf_outf);
ajPdbDel(&pdb);
ajStrDel(&ccf_this);
} /* End of main application loop. */
ajFmtPrint("STRIDE failures: %d\n", nostride);
ajFmtPrint("NACCESS failures: %d\n", nonaccess);
ajFmtPrintF(errf, "\n\nSTRIDE failures: %d\nNACCESS failures: %d\n",
nostride, nonaccess);
ajListFree(&ccfin);
ajDirDel(&pdbin);
ajStrDel(&pdbprefix);
ajStrDel(&pdb_name);
ajDiroutDel(&ccfout);
ajStrDel(&res);
ajStrDel(&res_num);
ajStrDel(&randomname);
ajStrDel(&line);
ajStrDel(&naccess_str);
ajStrDel(&exec);
ajStrDel(&syscmd);
ajFileClose(&errf);
if(ajStrGetCharFirst(*mode) != '2')
ajFileClose(&serrf);
if(ajStrGetCharFirst(*mode) != '1')
ajFileClose(&nerrf);
ajStrDel(&mode[0]);
AJFREE(mode);
ajExit();
return 0;
}
int main(int argc, char **argv)
{
ajint famn = 0; /* Counter for the families. */
ajint nset = 0; /* No. entries in family. */
ajint last_nodeid = 0; /* SCOP Sunid of last family that was
processed. */
AjPStr last_node = NULL; /* Last family that was processed. */
AjPStr exec = NULL; /* The UNIX command line to be executed. */
AjPStr out = NULL; /* Name of stamp alignment file. */
AjPStr align = NULL; /* Name of sequence alignment file. */
AjPStr alignc = NULL; /* Name of structure alignment file. */
AjPStr log = NULL; /* Name of STAMP log file. */
AjPStr dom = NULL; /* Name of file containing single domain. */
AjPStr set = NULL; /* Name of file containing set of domains. */
AjPStr scan = NULL; /* Name of temp. file used by STAMP. */
AjPStr sort = NULL; /* Name of temp. file used by STAMP. */
AjPStr name = NULL; /* Base name of STAMP temp files. */
AjPStr pdbnames = NULL; /* Names of domain pdb files to be passed to
TCOFFEEE. */
AjPDir pdb = NULL; /* Path of domain coordinate files (pdb
format input). */
AjPDirout daf = NULL; /* Path of sequence alignment files for output. */
AjPDirout super = NULL; /* Path of structure alignment files for output. */
AjPDirout singlets = NULL; /* Path of FASTA singlet sequence files for output. */
AjPStr temp1 = NULL; /* A temporary string. */
AjPFile dcfin = NULL; /* File pointer for original Escop.dat file.*/
AjPFile domf = NULL; /* File pointer for single domain file. */
AjPFile setf = NULL; /* File pointer for domain set file. */
AjPFile logf = NULL; /* Log file. */
AjPDomain domain = NULL; /* Pointer to domain structure. */
AjPDomain prevdomain = NULL; /* Pointer to previous domain structure. */
ajint type = 0; /* Type of domain (ajSCOP or ajCATH) in the
DCF file. */
AjPStr *node = NULL; /* Node of alignment . */
ajint noden = 0; /*1: Class (SCOP), 2: Fold (SCOP) etc, see
ACD file. */
AjPStr *mode = NULL; /* Mode of operation from acd*/
ajint moden = 0; /* Program mode, 1: MODE_STAMP, 2: MODE_TCOFFEE (not
yet implemented). */
AjBool keepsinglets= ajFalse; /*Whether to retain sequences of singlet families
and write them to an output file. */
AjPStr temp = NULL; /* A temporary string. */
AjPStr cmd = NULL; /* The command line to execute t-coffee. */
/* Initialise strings etc*/
last_node = ajStrNew();
exec = ajStrNew();
out = ajStrNew();
align = ajStrNew();
alignc = ajStrNew();
log = ajStrNew();
dom = ajStrNew();
set = ajStrNew();
scan = ajStrNew();
sort = ajStrNew();
name = ajStrNew();
temp = ajStrNew();
temp1 = ajStrNew();
cmd = ajStrNew();
pdbnames = ajStrNew();
/* Read data from acd. */
embInitPV("domainalign",argc,argv,"DOMALIGN",VERSION);
dcfin = ajAcdGetInfile("dcfinfile");
pdb = ajAcdGetDirectory("pdbdir");
daf = ajAcdGetOutdir("dafoutdir");
super = ajAcdGetOutdir("superoutdir");
singlets = ajAcdGetOutdir("singletsoutdir");
node = ajAcdGetList("node");
mode = ajAcdGetList("mode");
keepsinglets = ajAcdGetToggle("keepsinglets");
logf = ajAcdGetOutfile("logfile");
/* Convert the selected node and mode to an integer. */
if(!(ajStrToInt(node[0], &noden)))
ajFatal("Could not parse ACD node option");
if(!(ajStrToInt(mode[0], &moden)))
ajFatal("Could not parse ACD node option");
/* Initialise random number generator for naming of temp. files. */
ajRandomSeed();
ajFilenameSetTempname(&name);
/* Create names for temp. files. */
ajStrAssignS(&log, name);
ajStrAppendC(&log, ".log");
ajStrAssignS(&dom, name);
ajStrAppendC(&dom, ".dom");
ajStrAssignS(&set, name);
ajStrAppendC(&set, ".set");
ajStrAssignS(&scan, name);
ajStrAppendC(&scan, ".scan");
ajStrAssignS(&sort, name);
ajStrAppendC(&sort, ".sort");
ajStrAssignS(&out, name);
ajStrAppendC(&out, ".out");
/* Initialise last_node with something that is not in SCOP. */
ajStrAssignC(&last_node,"!!!!!");
/* Open STAMP domain set file. */
if(moden == MODE_STAMP)
{
if(!(setf=ajFileNewOutNameS(set)))
ajFatal("Could not open domain set file\n");
}
/* Get domain type. */
type = ajDomainDCFType(dcfin);
/* Start of main application loop. */
while((domain=(ajDomainReadCNew(dcfin, "*", type))))
{
/* A new family. */
if(((domain->Type == ajSCOP) &&
(((noden==1) && (last_nodeid != domain->Scop->Sunid_Class)) ||
((noden==2) && (last_nodeid != domain->Scop->Sunid_Fold)) ||
((noden==3) && (last_nodeid != domain->Scop->Sunid_Superfamily))||
((noden==4) && (last_nodeid != domain->Scop->Sunid_Family)))) ||
((domain->Type == ajCATH) &&
(((noden==5) && (last_nodeid != domain->Cath->Class_Id)) ||
((noden==6) && (last_nodeid != domain->Cath->Arch_Id)) ||
((noden==7) && (last_nodeid != domain->Cath->Topology_Id)) ||
((noden==8) && (last_nodeid != domain->Cath->Superfamily_Id)) ||
((noden==9) && (last_nodeid != domain->Cath->Family_Id)))))
{
/* If we have done the first family. */
if(famn)
{
/* Create the output file for the alignment - the name will
be the same as the Sunid for the DOMAIN family. */
domainalign_writeid(prevdomain, noden, daf, super,
&align, &alignc);
if(moden == MODE_STAMP)
{
/* Close domain set file. */
ajFileClose(&setf);
/* Call STAMP. */
/* Family with 2 or more entries. */
if(nset > 1)
{
domainalign_stamp(prevdomain,
domain,
daf,
super,
singlets,
align,
alignc,
dom,
name,
set,
scan,
sort,
log,
out,
keepsinglets,
moden,
noden,
nset,
logf);
}
else if(keepsinglets) /* Singlet family. */
domainalign_keepsinglets(prevdomain, noden,
singlets, logf);
/* Open STAMP domain set file. */
if(!(setf=ajFileNewOutNameS(set)))
ajFatal("Could not open domain set file\n");
}
else
{
/* Call TCOFEE. */
if(nset > 1)
domainalign_tcoffee(prevdomain, out, align,
alignc, pdbnames, noden, logf);
else if(keepsinglets) /* Singlet family. */
domainalign_keepsinglets(prevdomain, noden,
singlets, logf);
}
/* Set the number of members of the new family to zero. */
nset = 0;
/* Clear TCOFFEE argument. */
ajStrSetClear(&pdbnames);
}
/* Open, write and close STAMP domain file. */
if(moden == MODE_STAMP)
{
if(!(domf=ajFileNewOutNameS(dom)))
ajFatal("Could not open domain file\n");
ajStrAssignS(&temp, ajDomainGetId(domain));
ajStrFmtLower(&temp);
ajFmtPrintF(domf, "%S %S { ALL }\n", temp, temp);
ajFileClose(&domf);
}
/* Copy current family name to last_node. */
domainalign_writelast(domain, noden, &last_node, &last_nodeid);
/* Copy current domain pointer to prevdomain. */
ajDomainDel(&prevdomain);
prevdomain=NULL;
ajDomainCopy(&prevdomain, domain);
/* Increment family counter. */
famn++;
}
ajStrAssignS(&temp, ajDomainGetId(domain));
ajStrFmtLower(&temp);
/* Write STAMP domain set file. */
if(moden == MODE_STAMP)
ajFmtPrintF(setf, "%S %S { ALL }\n", temp, temp);
/* Write TCOFFEE argument. */
else
{
ajStrAppendS(&pdbnames, ajDirGetPath(pdb));
ajStrAppendS(&pdbnames, temp);
ajStrAppendC(&pdbnames, ".");
ajStrAppendS(&pdbnames, ajDirGetExt(pdb));
ajStrAppendC(&pdbnames, " ");
}
ajDomainDel(&domain);
/* Increment number of members in family. */
nset++;
}
/* End of main application loop. */
domain=prevdomain;
ajFmtPrint("\nProcessing node %d\n", last_nodeid);
/* Create the output file for the alignment - the name will
be the same as the Sunid for the DOMAIN family. */
domainalign_writeid(prevdomain, noden, daf, super, &align, &alignc);
/* Code to process last family. */
if(moden == MODE_STAMP)
{
/*Close domain set file. */
ajFileClose(&setf);
/* ajFmtPrint("\n***** SECOND CALL\n");. */
if(nset > 1)
{
domainalign_stamp(prevdomain,
domain,
daf,
super,
singlets,
align,
alignc,
dom,
name,
set,
scan,
sort,
log,
out,
keepsinglets,
moden,
noden,
nset,
logf);
}
else if(keepsinglets) /* Singlet family. */
domainalign_keepsinglets(prevdomain, noden, singlets, logf);
}
else
{
/* Call TCOFEE. */
if(nset > 1)
domainalign_tcoffee(prevdomain, out, align, alignc,
pdbnames, noden, logf);
else if(keepsinglets) /* Singlet family. */
domainalign_keepsinglets(prevdomain, noden, singlets, logf);
}
/* Remove all temporary files. */
ajSysFileUnlinkS(log);
ajSysFileUnlinkS(dom);
ajSysFileUnlinkS(set);
ajSysFileUnlinkS(scan);
ajSysFileUnlinkS(sort);
ajSysFileUnlinkS(out);
ajStrAssignS(&temp, name);
ajStrAppendC(&temp, ".mat");
ajSysFileUnlinkS(temp);
/* Tidy up*/
ajDomainDel(&domain);
ajFileClose(&dcfin);
ajStrDel(&last_node);
ajStrDel(&exec);
ajStrDel(&log);
ajStrDel(&dom);
ajStrDel(&set);
ajStrDel(&scan);
ajStrDel(&sort);
ajStrDel(&name);
ajStrDel(&out);
ajStrDel(&align);
ajStrDel(&alignc);
ajStrDel(&pdbnames);
ajDirDel(&pdb);
ajDiroutDel(&daf);
ajDiroutDel(&super);
ajDiroutDel(&singlets);
ajStrDel(&temp);
ajStrDel(&temp1);
ajStrDel(&node[0]);
AJFREE(node);
ajStrDel(&mode[0]);
AJFREE(mode);
ajFileClose(&logf);
ajExit();
return 0;
}
/* @prog seqnr **************************************************************
**
** Removes redundancy from DHF files (domain hits files) or other files of
** sequences.
**
****************************************************************************/
int main(int argc, char **argv)
{
/* Variable declarations */
AjPList in = NULL; /* Names of domain hits files (input). */
AjPStr inname = NULL; /* Full name of the current DHF file. */
AjPFile inf = NULL; /* Current DHF file. */
EmbPHitlist infhits = NULL; /* Hitlist from DHF file */
AjBool dosing = ajFalse; /* Filter using singlet sequences. */
AjPDir singlets = NULL; /* Singlets (input). */
AjBool dosets = ajFalse; /* Filter using sets of sequences. */
AjPDir insets = NULL; /* Sets (input). */
AjPStr mode = NULL; /* Mode of operation */
ajint moden = 0; /* Mode 1: single threshold for redundancy
removal, 2: lower and upper thresholds
for redundancy removal. */
float thresh = 0.0; /* Threshold for non-redundancy. */
float threshlow = 0.0; /* Threshold (lower limit). */
float threshup = 0.0; /* Threshold (upper limit). */
AjPMatrixf matrix = NULL; /* Substitution matrix. */
float gapopen = 0.0; /* Gap insertion penalty. */
float gapextend = 0.0; /* Gap extension penalty. */
AjPDirout out = NULL; /* Domain hits files (output). */
AjPFile outf = NULL; /* Current DHF file (output). */
AjBool dored = ajFalse; /* True if redundant hits are output. */
AjPDirout outred = NULL; /* DHF files for redundant hits (output).*/
AjPFile redf = NULL; /* Current DHF file redundancy (output). */
AjPStr outname = NULL; /* Name of output file (re-used). */
AjPFile logf = NULL; /* Log file pointer. */
AjBool ok = ajFalse; /* Housekeeping. */
AjPSeqset seqset = NULL; /* Seqset (re-used). */
AjPSeqin seqin = NULL; /* Seqin (re-used). */
AjPList seq_list = NULL; /* Main list for redundancy removal. */
EmbPDmxNrseq seq_tmp = NULL; /* Temp. pointer for making seq_list. */
ajint seq_siz = 0; /* Size of seq_list. */
AjPUint keep = NULL; /* 1: Sequence in seq_list was classed as
non-redundant, 0: redundant. */
AjPUint nokeep = NULL; /* Inversion of keep array. */
ajint nseqnr = 0; /* No. non-redundant seqs. in seq_list. */
AjPStr filtername= NULL; /* Name of filter file (re-used). */
AjPFile filterf = NULL; /* Current filter file. */
EmbPHitlist hitlist = NULL; /* Hitlist from input file (re-used). */
AjPScopalg scopalg = NULL; /* Scopalg from input file. */
ajint x = 0; /* Housekeeping. */
/* Read data from acd. */
embInitPV("seqnr",argc,argv,"DOMSEARCH",VERSION);
in = ajAcdGetDirlist("dhfinpath");
dosing = ajAcdGetToggle("dosing");
singlets = ajAcdGetDirectory("singletsdir");
dosets = ajAcdGetToggle("dosets");
insets = ajAcdGetDirectory("insetsdir");
mode = ajAcdGetListSingle("mode");
thresh = ajAcdGetFloat("thresh");
threshlow = ajAcdGetFloat("threshlow");
threshup = ajAcdGetFloat("threshup");
matrix = ajAcdGetMatrixf("matrix");
gapopen = ajAcdGetFloat("gapopen");
gapextend = ajAcdGetFloat("gapextend");
out = ajAcdGetOutdir("dhfoutdir");
dored = ajAcdGetToggle("dored");
outred = ajAcdGetOutdir("redoutdir");
logf = ajAcdGetOutfile("logfile");
/* Housekeeping. */
filtername = ajStrNew();
outname = ajStrNew();
if(!(ajStrToInt(mode, &moden)))
ajFatal("Could not parse ACD node option");
/* Process each DHF (input) in turn. */
while(ajListPop(in,(void **)&inname))
{
ajFmtPrint("Processing %S\n", inname);
ajFmtPrintF(logf, "//\n%S\n", inname);
seq_list = ajListNew();
keep = ajUintNew();
nokeep = ajUintNew();
/**********************************/
/* Open DHF file */
/**********************************/
if((inf = ajFileNewInNameS(inname)) == NULL)
ajFatal("Could not open DHF file %S", inname);
/* Read DHF file. */
ok = ajFalse;
if(!(infhits = embHitlistReadFasta(inf)))
{
ajWarn("embHitlistReadFasta call failed in seqnr");
ajFmtPrintF(logf, "embHitlistReadFasta call failed in seqnr\n");
/* Read sequence set instead. */
seqset = ajSeqsetNew();
seqin = ajSeqinNew();
ajSeqinUsa(&seqin, inname);
if(!(ajSeqsetRead(seqset, seqin)))
ajFatal("SeqsetRead failed in seqsearch_psialigned");
if(ajSeqsetGetSize(seqset))
ok = ajTrue;
}
else
if(infhits->N)
ok = ajTrue;
/* Close DHF file. */
ajFileClose(&inf);
/* Process empty DHF files (should never occur). */
if(!ok)
{
ajWarn("Empty input file %S\n", inname);
ajFmtPrintF(logf, "Empty input file %S\n", inname);
if(infhits)
embHitlistDel(&infhits);
if(seqset)
ajSeqsetDel(&seqset);
if(seqin)
ajSeqinDel(&seqin);
continue;
}
/* 1. Create list of sequences from the main input directory.. */
if(infhits)
{
for(x=0; x<infhits->N; x++)
{
AJNEW0(seq_tmp);
seq_tmp->Seq = ajSeqNew();
ajStrAssignS(&seq_tmp->Seq->Acc,infhits->hits[x]->Acc);
ajStrAssignS(&seq_tmp->Seq->Seq,infhits->hits[x]->Seq);
ajListPushAppend(seq_list,seq_tmp);
}
}
else
{
for(x=0;x<ajSeqsetGetSize(seqset);x++)
{
AJNEW0(seq_tmp);
seq_tmp->Seq = ajSeqNew();
ajStrAssignS(&seq_tmp->Seq->Acc, ajSeqsetGetseqAccS(seqset, x));
ajStrAssignS(&seq_tmp->Seq->Seq, ajSeqsetGetseqSeqS(seqset, x));
ajListPushAppend(seq_list,seq_tmp);
}
ajSeqsetDel(&seqset);
ajSeqinDel(&seqin);
}
/**********************************/
/* Open singlets filter file */
/**********************************/
if(dosing)
{
/* Open singlets file. */
ajStrAssignS(&filtername, inname);
ajFilenameTrimPathExt(&filtername);
ajStrInsertS(&filtername, 0, ajDirGetPath(singlets));
ajStrAppendC(&filtername, ".");
ajStrAppendS(&filtername, ajDirGetExt(singlets));
if((filterf = ajFileNewInNameS(filtername)) == NULL)
{
ajWarn("Could not open DHF file %S",
filtername);
ajFmtPrint("Could not open singlets filter file %S",
filtername);
}
else
{
/* Read DHF file. */
ok = ajFalse;
if(!(hitlist = embHitlistReadFasta(filterf)))
{
ajWarn("embHitlistReadFasta call failed in seqnr");
ajFmtPrintF(logf,
"embHitlistReadFasta call failed in seqnr\n");
/* Read sequence set instead. */
seqset = ajSeqsetNew();
seqin = ajSeqinNew();
ajSeqinUsa(&seqin, inname);
if(!(ajSeqsetRead(seqset, seqin)))
ajFatal("SeqsetRead failed in seqnr");
if(ajSeqsetGetSize(seqset))
ok = ajTrue;
}
else
if(hitlist->N)
ok = ajTrue;
/* Close DHF file. */
ajFileClose(&filterf);
/* Process empty DHF files (should never occur). */
if(!ok)
{
ajWarn("Empty singlets filter file %S\n", filtername);
ajFmtPrintF(logf, "Empty singlets filter file %S\n",
filtername);
/* No continue this time. */
}
/* 2. Add sequences from filter directories to List but mark
them up (they are considered in the redundancy calculation
but never appear in the output files). */
if(hitlist)
{
for(x=0; x<hitlist->N; x++)
{
AJNEW0(seq_tmp);
seq_tmp->Seq = ajSeqNew();
seq_tmp->Garbage = ajTrue;
ajStrAssignS(&seq_tmp->Seq->Acc,hitlist->hits[x]->Acc);
ajStrAssignS(&seq_tmp->Seq->Seq,hitlist->hits[x]->Seq);
ajListPushAppend(seq_list,seq_tmp);
}
embHitlistDel(&hitlist);
}
else
{
for(x=0;x<ajSeqsetGetSize(seqset);x++)
{
AJNEW0(seq_tmp);
seq_tmp->Seq = ajSeqNew();
seq_tmp->Garbage = ajTrue;
ajStrAssignS(&seq_tmp->Seq->Acc,
ajSeqsetGetseqAccS(seqset, x));
ajStrAssignS(&seq_tmp->Seq->Seq,
ajSeqsetGetseqSeqS(seqset, x));
ajListPushAppend(seq_list,seq_tmp);
}
ajSeqsetDel(&seqset);
ajSeqinDel(&seqin);
}
}
}
/**********************************/
/* Open sets filter file */
/**********************************/
if(dosets)
{
/* Open sets file. */
ajStrAssignS(&filtername, inname);
ajFilenameTrimPathExt(&filtername);
ajStrInsertS(&filtername, 0, ajDirGetPath(insets));
ajStrAppendC(&filtername, ".");
ajStrAppendS(&filtername, ajDirGetExt(insets));
if((filterf = ajFileNewInNameS(filtername)) == NULL)
{
ajWarn("Could not open DAF file %S", filtername);
ajFmtPrint("Could not open sets filter file %S", filtername);
}
else
{
/* Read DAF file. */
ok = ajFalse;
if(!(ajDmxScopalgRead(filterf, &scopalg)))
{
ajWarn("ajDmxScopalgRead call failed in seqnr");
ajFmtPrintF(logf,
"ajDmxScopalgRead call failed in seqnr\n");
/* Read sequence set instead. */
seqset = ajSeqsetNew();
seqin = ajSeqinNew();
ajSeqinUsa(&seqin, inname);
if(!(ajSeqsetRead(seqset, seqin)))
ajFatal("SeqsetRead failed in seqnr");
if(ajSeqsetGetSize(seqset))
ok = ajTrue;
}
else
if(scopalg->N)
ok = ajTrue;
/* Close DHF file. */
ajFileClose(&filterf);
/* Process empty DHF files (should never occur). */
if(!ok)
{
ajWarn("Empty sets filter file %S\n",
filtername);
ajFmtPrintF(logf, "Empty sets filter file %S\n",
filtername);
/* No continue this time. */
}
/* 2. Add sequences from filter directories to List but mark
them up (they are considered in the redundancy calculation
but never appear in the output files).. */
if(scopalg)
{
for(x=0; x<scopalg->N; x++)
{
AJNEW0(seq_tmp);
seq_tmp->Seq = ajSeqNew();
seq_tmp->Garbage = ajTrue;
ajStrAssignS(&seq_tmp->Seq->Acc,scopalg->Codes[x]);
ajStrAssignS(&seq_tmp->Seq->Seq,scopalg->Seqs[x]);
/* Remove gap char's & whitespace. */
ajStrRemoveGap(&seq_tmp->Seq->Seq);
ajListPushAppend(seq_list,seq_tmp);
}
ajDmxScopalgDel(&scopalg);
}
else
{
for(x=0;x<ajSeqsetGetSize(seqset);x++)
{
AJNEW0(seq_tmp);
seq_tmp->Seq = ajSeqNew();
seq_tmp->Garbage = ajTrue;
ajStrAssignS(&seq_tmp->Seq->Acc,
ajSeqsetGetseqAccS(seqset, x));
ajStrAssignS(&seq_tmp->Seq->Seq,
ajSeqsetGetseqSeqS(seqset, x));
ajListPushAppend(seq_list,seq_tmp);
}
ajSeqsetDel(&seqset);
ajSeqinDel(&seqin);
}
}
}
/* 4. Identify redundant domains.. */
if(moden == 1)
{
if((!embDmxSeqNR(seq_list, &keep, &nseqnr, matrix, gapopen,
gapextend, thresh, ajTrue)))
ajFatal("embDmxSeqNR failure in seqnr");
}
else
{
if((!embDmxSeqNRRange(seq_list, &keep, &nseqnr, matrix, gapopen,
gapextend, threshlow, threshup, ajTrue)))
ajFatal("embDmxSeqNR failure in seqnr");
}
seq_siz = ajListGetLength(seq_list);
for(x=0; x<seq_siz; x++)
if(ajUintGet(keep, x) == 1)
ajUintPut(&nokeep, x, 0);
else
ajUintPut(&nokeep, x, 1);
/* Create output files. */
ajStrAssignS(&outname, inname);
ajFilenameTrimPathExt(&outname);
outf = ajFileNewOutNameDirS(outname, out);
if(dored)
redf = ajFileNewOutNameDirS(outname, outred);
/* 5. Write non-redundant domains to main output directory.
6. If specified, write redundant domains to output directory. */
embHitlistWriteSubsetFasta(outf, infhits, keep);
if(dored)
embHitlistWriteSubsetFasta(redf, infhits, nokeep);
embHitlistDel(&infhits);
ajFileClose(&outf);
ajFileClose(&redf);
ajStrDel(&inname);
while(ajListPop(seq_list, (void **) &seq_tmp))
{
ajSeqDel(&seq_tmp->Seq);
AJFREE(seq_tmp);
}
ajListFree(&seq_list);
ajUintDel(&keep);
ajUintDel(&nokeep);
}
/* Tidy up. */
ajListFree(&in);
if(singlets)
ajDirDel(&singlets);
if(insets)
ajDirDel(&insets);
ajDiroutDel(&out);
if(outred)
ajDiroutDel(&outred);
ajFileClose(&logf);
ajMatrixfDel(&matrix);
ajStrDel(&filtername);
ajStrDel(&outname);
ajStrDel(&mode);
embExit();
return 0;
}