void strip_out_standard_options(int * argc,char ** argv,void (*show_help)(FILE * ofp),void (*show_version)(FILE * ofp))
{
char * errlog;
if( (strip_out_boolean_argument(argc,argv,"help")) == TRUE ) {
(*show_help)(stdout);
exit(1);
}
if( (strip_out_boolean_argument(argc,argv,"version")) == TRUE ) {
(*show_version)(stdout);
exit(1);
}
if( (strip_out_boolean_argument(argc,argv,"silent")) == TRUE ) {
erroroff(REPORT);
erroroff(INFO);
erroroff(WARNING);
}
if( (strip_out_boolean_argument(argc,argv,"quiet")) == TRUE ) {
erroroff(REPORT);
erroroff(INFO);
}
if( (strip_out_boolean_argument(argc,argv,"erroroffstd")) == TRUE ) {
errorstderroff(WARNING);
}
if( (errlog=strip_out_assigned_argument(argc,argv,"errlog")) != NULL ) {
if( add_log_filename(errlog) == FALSE ) {
fatal("Could not use %s as a error log file\n",errlog);
} else {
warn("Logging errors to %s as well as stderr",errlog);
errorlogon(WARNING);
}
}
}
int main(int argc,char ** argv)
{
DPRunImpl * dpri = NULL;
ScanWiseHSPImpl * scani = NULL;
HSP2HitListImpl * hsp2hiti = NULL;
HitListOutputImpl * hloi = NULL;
ProteinIndexConstructor * pic = NULL;
HSPScanInterface * hsi;
HSPScanInterfacePara * para;
SearchStatInterface * ssi;
SearchStatInterface * ssl;
SeqLookupLoadPara * slp;
HSPset2HitPairPara * hsp2hit;
CompMat * mat;
SequenceDB * db;
Sequence * seq;
int ret;
int i;
int effective_db_size = 300000;
int kk;
int count = 0;
LinearHSPmanager * lm;
HitList * hl;
boolean use_mott = 1;
boolean trunc_best_hsp = 0;
boolean verbose = 0;
static struct rusage use;
struct timeval t0, t1;
gettimeofday(&t0, NULL);
dpri = new_DPRunImpl_from_argv(&argc,argv);
dpri->memory = DPIM_Explicit;
scani = new_ScanWiseHSPImpl_from_argv(&argc,argv);
hsp2hiti = new_HSP2HitListImpl_from_argv(&argc,argv);
hloi = new_HitListOutputImpl_from_argv(&argc,argv);
slp = new_SeqLookupLoadPara_from_argv(&argc,argv);
pic = new_ProteinIndexConstructor_from_argv(&argc,argv);
hsp2hit = new_HSPset2HitPairPara_from_argv(&argc,argv);
para = new_HSPScanInterfacePara_from_argv(&argc,argv);
verbose = strip_out_boolean_argument(&argc,argv,"verbose") ;
strip_out_boolean_def_argument(&argc,argv,"mott",&use_mott);
strip_out_boolean_def_argument(&argc,argv,"besthsp",&trunc_best_hsp);
strip_out_integer_argument(&argc,argv,"dbsize",&effective_db_size);
#ifdef SCAN_CORBA
sorb = get_Wise2Corba_Singleton(&argc,argv,"orbit-local-orb");
#endif
if( dpri == NULL ) {
fatal("Unable to build DPRun implementation. Bad arguments");
}
strip_out_standard_options(&argc,argv,show_help,show_version);
if( argc != 2 ) {
show_help(stdout);
exit(12);
}
/* ugly, but we don't want to bounce matrices around the network... */
mat = read_Blast_file_CompMat("BLOSUM62.bla");
erroroff(REPORT);
hsi = new_HSPScanInterface_from_ScanWiseHSPImpl(scani,pic,slp);
ssi = new_Mott_SearchStatInterface();
ssl = new_lookup_SearchStatInterface(40,2.3);
if( verbose ) {
info("contacted database");
}
db = single_fasta_SequenceDB(argv[1]);
if( db == NULL ) {
fatal("Could not open sequence db...\n");
}
for(seq = init_SequenceDB(db,&ret); seq != NULL;seq = get_next_SequenceDB(db) ) {
count++;
for(i=0;i<seq->len;i++) {
if( !isalpha(seq->seq[i]) ) {
fatal("Sequence position %d [%c] is not valid",i,seq->seq[i]);
}
seq->seq[i] = toupper(seq->seq[i]);
}
info("Processing %s",seq->name);
getrusage(RUSAGE_SELF,&use);
/* info("Before query %s %.3fu %.3fs\n", seq->name,
use.ru_utime.tv_sec + use.ru_utime.tv_usec*MICROSECOND,
use.ru_stime.tv_sec + use.ru_stime.tv_usec*MICROSECOND
);
*/
lm = (*hsi->scan_query)(hsi->data,seq,para);
fprintf(stderr,"Got linear manager is %d entries\n",lm->len);
if( lm->mat == NULL ) {
lm->mat = hard_link_CompMat(mat);
}
getrusage(RUSAGE_SELF,&use);
/*
info("After query %s %.3fu %.3fs\n", seq->name,
use.ru_utime.tv_sec + use.ru_utime.tv_usec*MICROSECOND,
use.ru_stime.tv_sec + use.ru_stime.tv_usec*MICROSECOND
);
*/
sort_LinearHSPmanager(lm,compare_HSPset_score);
if( trunc_best_hsp == 1 ) {
for(kk=1;kk<lm->len;kk++) {
free_HSPset(lm->set[kk]);
lm->set[kk] = NULL;
}
lm->len = 1;
}
getrusage(RUSAGE_SELF,&use);
/*
info("After sort %s %.3fu %.3fs\n", seq->name,
use.ru_utime.tv_sec + use.ru_utime.tv_usec*MICROSECOND,
use.ru_stime.tv_sec + use.ru_stime.tv_usec*MICROSECOND
);
*/
hl = HitList_from_HSP_HSP2HitListImpl(hsp2hiti,lm,dpri,hsp2hit);
getrusage(RUSAGE_SELF,&use);
/*
info("After conversion %s %.3fu %.3fs\n", seq->name,
use.ru_utime.tv_sec + use.ru_utime.tv_usec*MICROSECOND,
use.ru_stime.tv_sec + use.ru_stime.tv_usec*MICROSECOND
);
*/
free_LinearHSPmanager(lm);
if( use_mott == 1 ) {
apply_SearchStat_to_HitList(hl,ssi,effective_db_size);
} else {
for(kk=0;kk<hl->len;kk++) {
hl->pair[kk]->bit_score = hl->pair[kk]->raw_score / 2.0;
}
}
sort_HitList_by_score(hl);
show_HitList_HitListOutputImpl(hloi,hl,stdout);
getrusage(RUSAGE_SELF,&use);
/*
info("After output %s %.3fu %.3fs\n", seq->name,
use.ru_utime.tv_sec + use.ru_utime.tv_usec*MICROSECOND,
use.ru_stime.tv_sec + use.ru_stime.tv_usec*MICROSECOND
);
*/
free_HitList(hl);
free_Sequence(seq);
}
free_DPRunImpl(dpri);
free_HSPScanInterface(hsi);
gettimeofday(&t1, NULL);
fprintf(stderr, "[client stats] queries, time (s): %d %f\n",
count,
(t1.tv_sec - t0.tv_sec) +
(t1.tv_usec - t0.tv_usec) * 1e-6);
return 0;
}