CompMat * read_Blast_CompMat(FILE * ifp)
{
char buffer[MAXLINE];
int alphabet[MAXLINE];
char * runner;
int len;
int linenum;
int row;
CompMat * out;
/***
Skip over # lines...
read first line: is alphabet ie
A R T G .... *
***/
while( fgets(buffer,MAXLINE,ifp) != NULL)
if( buffer[0] != '#')
break;
/** loop over line, getting letters: warn if longer than one, or not a letter **/
for(len=0,runner=strtok(buffer,spacestr);runner != NULL;runner=strtok(NULL,spacestr)) {
if( *runner == '*' )
break; /* end column */
if( !isalpha((int)*runner) || strlen(runner) > 1 ) {
warn("In read Blast comp mat, probably an error: trying to interpret [%s] as an amino acid",runner);
return NULL;
}
alphabet[len++] = toupper((int)*runner) -'A';
}
out = blank_CompMat();
linenum = 0;
/** get len lines, each line, get len numbers and put them away **/
while( fgets(buffer,MAXLINE,ifp) != NULL ) {
if( linenum >= len )
break;
for(runner=strtok(buffer,spacestr),row = 0;runner != NULL && row < len;runner=strtok(NULL,spacestr),row++) {
if( is_integer_string(runner,&out->comp[alphabet[linenum]][alphabet[row]]) == FALSE ) {
warn("In read Blast comp mat, probably an error: trying to interpret [%s] as a number ... continuing",runner);
}
}
linenum++;
}
return out;
}
boolean strip_out_integer_argument(int * argc,char ** argv,char * tag,int * value)
{
char * arg;
if( (arg = strip_out_assigned_argument(argc,argv,tag)) == NULL )
return FALSE;
if( is_integer_string(arg,value) == FALSE ) {
warn("Argument [%s] to [%s] is not an integer. Not changing the value [%d]",arg,tag,value);
return FALSE;
}
return TRUE;
}
DPImplementation * new_DPImplementation_from_argstr(int * argc,char ** argv)
{
DPImplementation * out;
char * temp;
out = DPImplementation_alloc();
if( (strip_out_boolean_argument(argc,argv,"pthreads")) == TRUE ) {
out->do_threads = TRUE;
}
if( (temp=strip_out_assigned_argument(argc,argv,"dbtrace")) != NULL ) {
if( is_integer_string(temp,&out->db_trace_level) == FALSE ) {
warn("%s is not an integer argument for dbtrace",temp);
}
}
if( strip_out_boolean_argument(argc,argv,"O") == TRUE ) {
out->largemem= TRUE;
/* other optimisations */
}
strip_out_boolean_def_argument(argc,argv,"largemem",&out->largemem);
strip_out_boolean_def_argument(argc,argv,"onemodel",&out->doone);
if( strip_out_boolean_argument(argc,argv,"prob") == TRUE ) {
out->doprob = TRUE;
}
if( strip_out_boolean_argument(argc,argv,"g") == TRUE ) {
out->dydebug = TRUE;
}
if( (temp=strip_out_assigned_argument(argc,argv,"logsum")) != NULL ) {
out->calcfunc = stringalloc(temp);
} else {
out->calcfunc = stringalloc("Probability_logsum");
}
out->dycw = new_DycWarning_from_argstr(argc,argv);
/* fprintf(stderr,"And %d is extern warning",out->dycw->warn_extern);*/
return out;
}
DPRunImpl * new_DPRunImpl_from_argv(int * argc,char ** argv)
{
DPRunImpl * out;
char * temp;
out = DPRunImpl_alloc();
if( (temp = strip_out_assigned_argument(argc,argv,"dymem")) != NULL ) {
if( strcmp(temp,"explicit") == 0) {
out->memory = DPIM_Explicit;
} else if( strcmp(temp,"linear") == 0 ) {
out->memory = DPIM_Linear;
} else if( strcmp(temp,"default") == 0 ) {
out->memory = DPIM_Default;
} else {
warn("String [%s] for dynamic memory layout is not recognised",temp);
free_DPRunImpl(out);
return NULL;
}
}
if( (temp = strip_out_assigned_argument(argc,argv,"kbyte")) != NULL ) {
if( is_integer_string(temp,&out->kbyte_size) == FALSE ) {
warn("String [%s] for dynamic memory size is not recognised",temp);
free_DPRunImpl(out);
return NULL;
}
}
if(strip_out_boolean_argument(argc,argv,"dydebug") == TRUE ) {
out->debug = 1;
out->memory = DPIM_Explicit;
}
return out;
}
Gene * read_EMBL_feature_Gene(char * buffer,int maxlen,FILE * ifp)
{
Gene * gene;
Transcript * tr;
Translation * ts;
Exon * exon;
char * runner;
char * base;
char * next;
int i;
int exon_start[MAX_EMBL_EXON_PARSE];
int exon_end[MAX_EMBL_EXON_PARSE];
int number;
int exon_no = 0;
int isstart = 1;
int is_complement = 0;
int is_cds = 0;
int break_at_end = 0;
if( strstartcmp(buffer,"FT") != 0 ) {
warn("passed in a bad line [%s] to be used for feature table parsing",buffer);
return NULL;
}
if( (runner=strtok(buffer+2,spacestr)) == NULL ) {
warn("Bad embl feature line [%s]",buffer);
return NULL;
}
if( strcmp(runner,"CDS") != 0 && strcmp(runner,"mRNA") != 0 ) {
warn("passed in a feature line to read_EMBL_feature_Gene with a %s tag. This only handles CDS and mRNA tags",runner);
return NULL;
}
if( strcmp(runner,"CDS") == 0 ) {
is_cds = TRUE;
}
runner = strtok(NULL,spacestr);
if( runner == NULL ) {
warn("Bad embl feature line [%s]",buffer);
return NULL;
}
if( strstartcmp(runner,"complement") == 0 ) {
runner = strchr(runner,'(');
if( runner == NULL) {
warn("Could not find bracket on EMBL feature complement line");
return NULL;
}
is_complement = 1;
runner++;
}
if( strstartcmp(runner,"join") == 0 ) {
runner = strchr(runner,'(');
runner++;
} else if( isdigit((int)*runner) || *runner == '<' ) {
/** ok - starts with the numbers. We'll cope!**/
} else {
warn("Expecting a join statement, got a [%s]",runner);
return NULL;
}
/*** ok, now the major number loop ***/
for(;;) {
base= runner;
for(;*runner && *runner != ')' && *runner != '.' && *runner != ',' && *runner != '>' && !isspace((int)*runner);runner++)
;
/*fprintf(stderr,"Got a runner of %s\n ",runner); */
if( *runner == '\0' )
next = runner;
else next = runner+1;
if( *runner == ')' ) {
break_at_end = TRUE; /* out of reading exons */
}
*runner='\0';
if( strstartcmp(base,"complement(") == 0 ) {
is_complement = TRUE;
for(;*base != '(';base++)
;
base++;
break_at_end = FALSE; /* we found an bracket too early! */
}
if( is_integer_string(base,&number) == FALSE ) {
warn("Got a non integer [%s] in the middle of a join statement in EMBL parsing",runner);
return NULL;
}
/** put this number away **/
if( isstart ) {
exon_start[exon_no] = number;
isstart = 0;
} else {
exon_end[exon_no++] = number;
isstart = 1;
}
if( break_at_end == TRUE)
break;
for(runner=next;*runner && (*runner == '.' || isspace((int)*runner));runner++)
;
if( *runner == '\0' ) {
if( next_feature_tab_line(buffer,maxlen,ifp) == FALSE) {
warn("In the middle of getting a join statement, got a [%s]. Yuk!",buffer);
return NULL;
}
if( !isdigit((int)buffer[0]) && buffer[0] != '.' && buffer[0] != ',') {
/*** ok - sometimes people very boring end things in here ***/
/* warn("In the middle of getting a join statement, got a [%s]. Ugh!",buffer); */
break;
}
runner = buffer;
}
}
if( isstart == 0 ) {
warn("I have read an uneven number of start-end points in the exon thing. Yuk!");
return NULL;
}
/** runner should now be on bracket **/
if( is_complement == 1 ) {
/** ok . should be another bracket. Do we care? **/
}
gene = Gene_alloc_len(1);
tr = Transcript_alloc_len(exon_no);
add_Gene(gene,tr);
tr->parent = gene;
if( is_complement == 1 ) {
gene->start = exon_end[exon_no-1]-1;
gene->end = exon_start[0] -1;
for(i=exon_no -1;i >= 0;i--) {
exon = Exon_alloc();
exon->start = (gene->start+1) - exon_end[i];
exon->end = (gene->start+1) - exon_start[i] +1;
add_ex_Transcript(tr,exon);
}
} else {
gene->start = exon_start[0] -1;
gene->end = exon_end[exon_no-1] -1;
for(i=0;i<exon_no;i++) {
exon = Exon_alloc();
exon->start = exon_start[i] - (gene->start+1);
exon->end = exon_end[i] - (gene->start+1)+1;
add_ex_Transcript(tr,exon);
}
}
if( is_cds == TRUE ) {
ts = Translation_alloc();
ts->start = 0;
ts->end = length_Transcript(tr);
ts->parent = tr;
add_Transcript(tr,ts);
}
/*** read the rest of this feature ***/
while( next_feature_tab_line(buffer,maxlen,ifp) == TRUE)
;
return gene;
}
int main(int argc,char * argv[])
{
FailureType fail = 0 ;
FailureType should_fail_on = 0;
int i;
boolean doinfo = FALSE;
boolean noaddnumbers = FALSE;
MethodTypeSet * mts;
MethodTypeSet * cp;
boolean no_config_mts = FALSE;
int prot_level = 0;
int should_hard_link = 0;
boolean should_warn_undoc = FALSE;
char * prot_str;
char * runner;
char *config_dir=NULL;
char buffer[64]; /** really for removing files **/
char * telegraph;
APIpara api;
char * pack;
/** we no longer read in configs **/
mts = standard_dynamite_MethodTypeSet();
if( strip_out_boolean_argument(&argc,argv,"h") == TRUE
|| strip_out_boolean_argument(&argc,argv,"u") == TRUE /* arve */
|| argc == 1 ) {
show_usage(stdout);
exit(1);
}
noaddnumbers = strip_out_boolean_argument(&argc,argv,"m");
doinfo = strip_out_boolean_argument(&argc,argv,"i");
no_config_mts = strip_out_boolean_argument(&argc,argv,"U");
should_hard_link = strip_out_boolean_argument(&argc,argv,"l");
prot_str = strip_out_assigned_argument(&argc,argv,"P");
should_warn_undoc = strip_out_boolean_argument(&argc,argv,"D");
telegraph = strip_out_assigned_argument(&argc,argv,"tele");
pack = strip_out_assigned_argument(&argc,argv,"n");
api.xs_ext = NULL;
api.typemap_ext = NULL;
api.pod_ext = NULL;
api.c_extension_name = strip_out_assigned_argument(&argc,argv,"a");
api.t_extension_name = strip_out_assigned_argument(&argc,argv,"b");
api.pfdoc_ext = strip_out_assigned_argument(&argc,argv,"p");
api.xs_ext = strip_out_assigned_argument(&argc,argv,"x");
api.typemap_ext = strip_out_assigned_argument(&argc,argv,"tym");
api.all_callable = strip_out_boolean_argument(&argc,argv,"c");
api.make_perl = strip_out_boolean_argument(&argc,argv,"perl");
api.latex_ext = strip_out_assigned_argument(&argc,argv,"exttex");
api.make_latex = strip_out_boolean_argument(&argc,argv,"latex");
if( api.make_perl == TRUE) {
if( api.xs_ext == NULL ) {
api.xs_ext = ".xs";
}
if( api.typemap_ext == NULL ) {
api.typemap_ext = ".typemap";
}
if( api.pod_ext == NULL ) {
api.pod_ext = ".pod";
}
}
if( strip_out_boolean_argument(&argc,argv,"F") == TRUE) {
should_fail_on = FailureType_dyc_All;
}
/* do DPImplementation */
dpi = new_DPImplementation_from_argstr(&argc,argv);
if( prot_str != NULL ) {
if( is_integer_string(prot_str,&prot_level) == FALSE ) {
warn("Protection level %s is no integer!");
prot_level = 0;
}
}
/* Override/set WISECONFIGDIR on the cmdline. (arve) */
config_dir = strip_out_assigned_argument(&argc, argv, "I");
if (config_dir != NULL) {
set_config_dir(config_dir);
}
if( read_into_MethodTypeSet_filename(mts,"methods") == FALSE){
warn("You have no config file called 'methods'. This is bad news for dynamite matrices. I will attempt to compile, but you cannot use logical types. 'methods' should be either in the current directory, the $WISECONFIGDIR or your $WISEPERSONALDIR");
}
/*** ok,loop over and do it ***/
if( argc < 1 ) {
warn("You must have at least one dynamite source file to compile!");
show_usage(stdout);
exit(1);
}
if( telegraph != NULL ) {
tele_file= fopen(telegraph,"w");
}
for(i=1;i<argc;i++) {
if( mts != NULL)
cp = copy_MethodTypeSet(mts); /* actually very cheap */
if( do_a_file(argv[i],mts,FALSE,prot_level,should_hard_link,should_warn_undoc,noaddnumbers == TRUE ? FALSE : TRUE,pack,&api,&fail) == FALSE ) {
fatal("Terminated dyc one %d argument %s",i,argv[i]);
}
if( (should_fail_on == 01 && fail != 0) || (fail & should_fail_on) ) {
/*** remove files which fail ****/
/*** ugh this should be done better ***/
for(runner=argv[i]+strlen(argv[i]) - 1;runner > argv[i] && *runner != '.';runner--)
;
if( runner != argv[i] ) {
*runner = '\0';
sprintf(buffer,"%s.c",argv[i]);
if( remove_file(buffer) == FALSE ) {
warn("Could not remove file %s from filesystem",buffer);
}
sprintf(buffer,"%s.h",argv[i]);
if( remove_file(buffer) == FALSE ) {
warn("Could not remove file %s from filesystem",buffer);
}
}
/* else - well - something bad has happened */
fatal("Failed on file %s due to user defined fails",argv[i]);
}
if( mts != NULL ) {
free_MethodTypeSet(mts);
mts = cp;
}
}
free_MethodTypeSet(mts);
return 0;
}
boolean build_objects(void)
{
boolean ret = TRUE;
Protein * pro_temp;
Genomic * gen_temp;
FILE * ifp;
startend = threestatemodel_mode_from_string(startend_string);
if( startend == TSM_unknown ) {
warn("String %s was unable to converted into a start/end policy\n",startend_string);
ret = FALSE;
}
if( tstart_str != NULL ) {
if( is_integer_string(tstart_str,&tstart) == FALSE || tstart < 0) {
warn("Could not make %s out as target start",tstart);
ret = FALSE;
}
}
if( tend_str != NULL ) {
if( is_integer_string(tend_str,&tend) == FALSE || tend < 0) {
warn("Could not make %s out as target end",tend);
ret = FALSE;
}
}
if( is_integer_string(gap_str,&gap) == FALSE ) {
warn("Could not make %s out as gap penalty (must be integer at the moment)",gap_str);
ret = FALSE;
}
if( is_integer_string(ext_str,&ext) == FALSE ) {
warn("Could not make %s out as gap penalty (must be integer at the moment)",ext_str);
ret = FALSE;
}
if( is_embl == FALSE ) {
if( (gen = read_fasta_file_Genomic(dna_seq_file,length_of_N)) == NULL ) {
ret = FALSE;
warn("Could not read genomic sequence in %s",dna_seq_file);
gen = NULL;
}
} else {
embl = read_EMBL_GenomicRegion_file(dna_seq_file);
if( embl == NULL ) {
warn("Could not read genomic EMBL file in %s",dna_seq_file);
gen = NULL;
ret = FALSE;
} else {
gen = hard_link_Genomic(embl->genomic);
}
}
if( gen != NULL ) {
if( tstart != -1 || tend != -1 ) {
if( tstart == -1 )
tstart = 0;
if( tend == -1 )
tend = gen->baseseq->len;
gen_temp = truncate_Genomic(gen,tstart-1,tend);
if( gen_temp == NULL ){
ret = FALSE;
} else {
free_Genomic(gen);
gen = gen_temp;
}
} else {
/* no truncation required */
}
if( reverse == TRUE ) {
if( tstart > tend ) {
warn("You have already reversed the DNA by using %d - %d truncation. Re-reversing",tstart,tend);
}
gen_temp = reverse_complement_Genomic(gen);
free_Genomic(gen);
gen = gen_temp;
}
}
/*
* Can't truncate on GenomicRegion (for good reasons!).
* but we want only a section of the EMBL file to be used
*
* So... swap genomic now. Positions in EMBL are still valid,
* however - some genes will loose their sequence, which will be damaging. ;)
*/
if( is_embl ) {
free_Genomic(embl->genomic);
embl->genomic = hard_link_Genomic(gen); /* pointer could be dead anyway ;) */
}
if( target_abs == TRUE ) {
if( is_embl == TRUE ) {
warn("Sorry you can't both use absolute positioning and EMBL files as I can't cope with all the coordinate remapping. You'll have to convert to fasta.");
ret = FALSE;
}
gen->baseseq->offset = 1;
gen->baseseq->end = strlen(gen->baseseq->seq);
}
if( alg_str != NULL ) {
alg = gwrap_alg_type_from_string(alg_str);
} else {
if( use_tsm == TRUE ) {
alg_str = "623L";
} else {
alg_str = "623";
}
alg = gwrap_alg_type_from_string(alg_str);
}
if( qstart_str != NULL ) {
if( is_integer_string(qstart_str,&qstart) == FALSE || qstart < 0) {
warn("Could not make %s out as query start",qstart);
ret = FALSE;
}
}
if( qend_str != NULL ) {
if( is_integer_string(qend_str,&qend) == FALSE || qend < 0) {
warn("Could not make %s out as query end",qend);
ret = FALSE;
}
}
if( use_tsm == FALSE ) {
if( startend != TSM_default && startend != TSM_global && startend != TSM_local && startend != TSM_endbiased) {
warn("Proteins can only have local/global/endbias startend policies set, not %s",startend_string);
ret = FALSE;
}
if( (pro = read_fasta_file_Protein(protein_file)) == NULL ) {
ret = FALSE;
warn("Could not read Protein sequence in %s",protein_file);
} else {
if( qstart != -1 || qend != -1 ) {
if( qstart == -1 )
qstart = 0;
if( qend == -1 )
qend = pro->baseseq->len;
pro_temp = truncate_Protein(pro,qstart-1,qend);
if( pro_temp == NULL ){
ret = FALSE;
} else {
free_Protein(pro);
pro = pro_temp;
}
}
}
} else {
/** using a HMM **/
/*tsm = read_HMMer_1_7_ascii_file(hmm_file);*/
/*tsm = Wise2_read_ThreeStateModel_from_hmmer1_file(hmm_file);*/
tsm = HMMer2_read_ThreeStateModel(hmm_file);
if( tsm == NULL ) {
warn("Could not read hmm from %s\n",hmm_file);
ret = FALSE;
} else {
display_char_in_ThreeStateModel(tsm);
if( hmm_name != NULL ) {
if( tsm->name != NULL )
ckfree(tsm->name);
tsm->name = stringalloc(hmm_name);
}
if( tsm == NULL ) {
warn("Could not read %s as a hmm",hmm_file);
}
/** have to set start/end **/
set_startend_policy_ThreeStateModel(tsm,startend,30,0.1);
}
} /* end of else tsm != NULL */
if( main_block_str != NULL ) {
if( is_integer_string(main_block_str,&main_block) == FALSE ) {
warn("Could not get maximum main_block number %s",main_block_str);
ret = FALSE;
}
}
if( is_double_string(subs_string,&subs_error) == FALSE ) {
warn("Could not convert %s to a double",subs_error);
ret = FALSE;
}
if( is_double_string(indel_string,&indel_error) == FALSE ) {
warn("Could not convert %s to a double",indel_error);
ret = FALSE;
}
if( is_double_string(allN_string,&allN) == FALSE ) {
warn("Could not convert %s to a double",allN_string);
ret = FALSE;
}
if( strcmp(cfreq_string,"model") == 0 ) {
model_codon = TRUE;
} else if ( strcmp(cfreq_string,"flat") == 0 ) {
model_codon = FALSE;
} else {
warn("Cannot interpret [%s] as a codon modelling parameter\n",cfreq_string);
ret = FALSE;
}
if( strcmp(splice_string,"model") == 0 ) {
model_splice = TRUE;
} else if ( strcmp(splice_string,"flat") == 0 ) {
model_splice = FALSE;
gmp->use_gtag_splice = TRUE;
} else {
warn("Cannot interpret [%s] as a splice modelling parameter\n",splice_string);
ret = FALSE;
}
if( strcmp(null_string,"syn") == 0 ) {
use_syn = TRUE;
} else if ( strcmp(null_string,"flat") == 0 ) {
use_syn = FALSE;
} else {
warn("Cannot interpret [%s] as a null model string\n",null_string);
ret = FALSE;
}
if( strcmp(intron_string,"model") == 0 ) {
use_tied_model = FALSE;
} else if ( strcmp(intron_string,"tied") == 0 ) {
use_tied_model = TRUE;
} else {
warn("Cannot interpret [%s] as a intron tieing switch\n",intron_string);
ret = FALSE;
}
if( (rm = default_RandomModel()) == NULL) {
warn("Could not make default random model\n");
ret = FALSE;
}
if( use_new_stats == 0 ) {
if( (gf = read_GeneFrequency21_file(gene_file)) == NULL) {
ret = FALSE;
warn("Could not read a GeneFrequency file in %s",gene_file);
}
} else {
if( (gs = GeneStats_from_GeneModelParam(gmp)) == NULL ){
ret=FALSE;
warn("Could not read gene statistics in %s",new_gene_file);
}
} /* end of else using new gene stats */
if( (mat = read_Blast_file_CompMat(matrix_file)) == NULL) {
if( use_tsm == TRUE ) {
info("I could not read the Comparison matrix file in %s; however, you are using a HMM so it is not needed. Please set the WISECONFIGDIR or WISEPERSONALDIR variable correctly to prevent this message.",matrix_file);
} else {
warn("Could not read Comparison matrix file in %s",matrix_file);
ret = FALSE;
}
}
if( (ct = read_CodonTable_file(codon_file)) == NULL) {
ret = FALSE;
warn("Could not read codon table file in %s",codon_file);
}
if( (ofp = openfile(output_file,"W")) == NULL) {
warn("Could not open %s as an output file",output_file);
ret = FALSE;
}
rmd = RandomModelDNA_std();
return ret;
}
GeneStats * read_GeneStats(FILE * ifp)
{
char buffer[MAXLINE];
GeneStats * out;
SeqAlign * temp;
char **base;
char **brk;
out = GeneStats_alloc();
out->rnd = NULL;
while( fgets(buffer,MAXLINE,ifp) != NULL ) {
/* fprintf(stderr,"Reading (main loop) %s",buffer); */
if( buffer[0] == '#' )
continue;
if( buffer[0] == '%' && buffer[1] == '%' )
break;
if( strstartcmp(buffer,"splice5") == 0 ) {
base = brk = breakstring(buffer,spacestr);
if( *brk == NULL || *(brk+1) == NULL || is_integer_string(*(brk+1),&out->splice5_offset) == 0) {
warn("Cannot read splice5 offset - must be splice5 <number>");
return NULL;
}
ckfree(base);
temp = read_selex_SeqAlign(ifp);
if( temp == NULL ) {
warn("Could not read in selex alignment for splice5");
continue;
}
out->splice5 = temp;
continue;
}
if( strstartcmp(buffer,"splice3") == 0 ) {
base = brk = breakstring(buffer,spacestr);
if( *brk == NULL || *(brk+1) == NULL || is_integer_string(*(brk+1),&out->splice3_offset) == 0) {
warn("Cannot read splice3 offset - must be splice3 <number>");
return NULL;
}
ckfree(base);
temp = read_selex_SeqAlign(ifp);
if( temp == NULL ) {
warn("Could not read in selex alignment for splice5");
continue;
}
out->splice3 = temp;
continue;
}
if( strstartcmp(buffer,"intron_emission") == 0 ) {
if( fgets(buffer,MAXLINE,ifp) == NULL ) {
warn("Could not read in intron emission line");
break;
}
out->intron = get_genestat_emission(buffer);
if( fgets(buffer,MAXLINE,ifp) != NULL ) {
continue;
} else {
break;
}
}
if( strstartcmp(buffer,"polyp_emission") == 0 ) {
if( fgets(buffer,MAXLINE,ifp) == NULL ) {
warn("Could not read in polyp emission line");
break;
}
out->polyp = get_genestat_emission(buffer);
if( fgets(buffer,MAXLINE,ifp) != NULL ) {
continue;
} else {
break;
}
}
if( strstartcmp(buffer,"rnd_emission") == 0 ) {
if( fgets(buffer,MAXLINE,ifp) == NULL ) {
warn("Could not read in rnd emission line");
break;
}
out->rnd = get_genestat_emission(buffer);
if( fgets(buffer,MAXLINE,ifp) != NULL ) {
continue;
} else {
break;
}
}
if( strstartcmp(buffer,"rndcodon") == 0 ) {
if( read_codon_GeneStats(out->codon,buffer,ifp) == FALSE ) {
warn("Problem in reading codon line!");
}
continue;
}
if( isalpha(buffer[0]) ) {
warn("Could not read line %s in genestats reading\n",buffer);
}
}
assert(out);
assert(out->splice5);
assert(out->splice3);
return out;
}
boolean build_objects(void)
{
boolean ret = TRUE;
Protein * pro_temp;
SequenceDB * psdb;
startend = threestatemodel_mode_from_string(startend_string);
if( startend == TSM_unknown ) {
warn("String %s was unable to converted into a start/end policy\n",startend_string);
ret = FALSE;
}
if( use_single_dna == TRUE ) {
cdna = read_fasta_file_cDNA(dna_seq_file);
if( cdna == NULL ) {
warn("Could not open single dna sequence in %s",dna_seq_file);
ret = FALSE;
}
} else {
sdb = single_fasta_SequenceDB(dna_seq_file);
if( sdb == NULL ) {
warn("Could not build a sequence database on %s",dna_seq_file);
ret = FALSE;
}
}
rm = default_RandomModel();
if( (mat = read_Blast_file_CompMat(matrix_file)) == NULL) {
if( use_tsm == TRUE ) {
info("I could not read the Comparison matrix file in %s; however, you are using a HMM so it is not needed. Please set the WISECONFIGDIR or WISEPERSONALDIR variable correctly to prevent this message.",matrix_file);
} else {
warn("Could not read Comparison matrix file in %s",matrix_file);
ret = FALSE;
}
}
if( is_integer_string(gap_str,&gap) == FALSE ) {
warn("Could not get gap string number %s",gap_str);
ret = FALSE;
}
if( is_integer_string(ext_str,&ext) == FALSE ) {
warn("Could not get ext string number %s",ext_str);
ret = FALSE;
}
if( qstart_str != NULL ) {
if( is_integer_string(qstart_str,&qstart) == FALSE || qstart < 0) {
warn("Could not make %s out as query start",qstart);
ret = FALSE;
}
}
if( qend_str != NULL ) {
if( is_integer_string(qend_str,&qend) == FALSE || qend < 0) {
warn("Could not make %s out as query end",qend);
ret = FALSE;
}
}
if( aln_number_str != NULL ) {
if( is_integer_string(aln_number_str,&aln_number) == FALSE || aln_number < 0) {
warn("Weird aln number string %s...\n",aln_number_str);
ret = FALSE;
}
}
if( report_str != NULL ) {
if( is_integer_string(report_str,&report_stagger) == FALSE ) {
warn("Weird report stagger asked for %s",report_str);
ret = FALSE;
}
}
if( use_pfam1 == TRUE ) {
tsmdb = new_PfamHmmer1DB_ThreeStateDB(protein_file);
if( set_search_type_ThreeStateDB(tsmdb,startend_string) == FALSE) {
warn("Unable to set global/local switch on threestatedb");
ret = FALSE;
}
} else if ( use_pfam2 == TRUE ) {
tsmdb = HMMer2_ThreeStateDB(protein_file);
if( set_search_type_ThreeStateDB(tsmdb,startend_string) == FALSE) {
warn("Unable to set global/local switch on threestatedb");
ret = FALSE;
}
} else if ( use_tsm == TRUE) {
/** using a HMM **/
tsm = HMMer2_read_ThreeStateModel(protein_file);
if( tsm == NULL ) {
warn("Could not read hmm from %s\n",protein_file);
ret = FALSE;
} else {
display_char_in_ThreeStateModel(tsm);
if( hmm_name != NULL ) {
if( tsm->name != NULL )
ckfree(tsm->name);
tsm->name = stringalloc(hmm_name);
} else {
if( tsm->name == NULL ) {
tsm->name = stringalloc(protein_file);
}
}
/** have to set start/end **/
set_startend_policy_ThreeStateModel(tsm,startend,15,0.2);
tsmdb = new_single_ThreeStateDB(tsm,rm);
if( tsmdb == NULL ) {
warn("Could not build a threestatemodel database from a single tsm. Weird!");
ret = FALSE;
}
} /* end of else tsm != NULL */
} /* end of else is tsm */
else if( use_single_pro ) {
if( startend != TSM_default && startend != TSM_global && startend != TSM_local ) {
warn("Proteins can only have local/global startend policies set, not %s",startend_string);
ret = FALSE;
}
if( (pro = read_fasta_file_Protein(protein_file)) == NULL ) {
ret = FALSE;
warn("Could not read Protein sequence in %s",protein_file);
} else {
if( qstart != -1 || qend != -1 ) {
if( qstart == -1 )
qstart = 0;
if( qend == -1 )
qend = pro->baseseq->len;
pro_temp = truncate_Protein(pro,qstart-1,qend);
if( pro_temp == NULL ){
ret = FALSE;
} else {
free_Protein(pro);
pro = pro_temp;
}
}
if( startend == TSM_global)
tsm = global_ThreeStateModel_from_half_bit_Sequence(pro,mat,rm,-gap,-ext);
else
tsm = ThreeStateModel_from_half_bit_Sequence(pro,mat,rm,-gap,-ext);
if( tsm == NULL ) {
warn("Could not build ThreeStateModel from a single protein sequence...");
ret = FALSE;
} else {
tsmdb = new_single_ThreeStateDB(tsm,rm);
if( tsmdb == NULL ) {
warn("Could not build a threestatemodel database from a single tsm. Weird!");
ret = FALSE;
}
} /* end of could build a TSM */
} /* else is a real protein */
} /* end of else is single protein */
else if (use_db_pro == TRUE ) {
psdb = single_fasta_SequenceDB(protein_file);
tsmdb = new_proteindb_ThreeStateDB(psdb,mat,-gap,-ext);
free_SequenceDB(psdb);
}
else {
warn("No protein input file! Yikes!");
}
/***
if( use_tsm == FALSE ) {
} else {
****/
if( main_block_str != NULL ) {
if( is_integer_string(main_block_str,&main_block) == FALSE ) {
warn("Could not get maximum main_block number %s",main_block_str);
ret = FALSE;
}
}
if( evalue_search_str != NULL && is_double_string(evalue_search_str,&evalue_search_cutoff) == FALSE ) {
warn("Could not convert %s to a double",evalue_search_str);
ret = FALSE;
}
if( is_double_string(search_cutoff_str,&search_cutoff) == FALSE ) {
warn("Could not convert %s to a double",search_cutoff_str);
ret = FALSE;
}
if( is_double_string(subs_string,&subs_error) == FALSE ) {
warn("Could not convert %s to a double",subs_error);
ret = FALSE;
}
if( is_double_string(indel_string,&indel_error) == FALSE ) {
warn("Could not convert %s to a double",indel_error);
ret = FALSE;
}
if( is_double_string(allN_string,&allN) == FALSE ) {
warn("Could not convert %s to a double",allN_string);
ret = FALSE;
}
if( strcmp(null_string,"syn") == 0 ) {
use_syn = TRUE;
} else if ( strcmp(null_string,"flat") == 0 ) {
use_syn = FALSE;
} else {
warn("Cannot interpret [%s] as a null model string\n",null_string);
ret = FALSE;
}
if( alg_str != NULL ) {
alg = alg_estwrap_from_string(alg_str);
} else {
alg_str = "312";
alg = alg_estwrap_from_string(alg_str);
}
if( aln_alg_str != NULL ) {
aln_alg = alg_estwrap_from_string(aln_alg_str);
} else {
/* if it is a protein, don't loop */
if( use_single_pro == TRUE || use_db_pro == TRUE )
aln_alg_str = "333";
else
aln_alg_str = "333L";
aln_alg = alg_estwrap_from_string(aln_alg_str);
}
if( (rm = default_RandomModel()) == NULL) {
warn("Could not make default random model\n");
ret = FALSE;
}
if( (ct = read_CodonTable_file(codon_file)) == NULL) {
ret = FALSE;
warn("Could not read codon table file in %s",codon_file);
}
if( (ofp = openfile(output_file,"W")) == NULL) {
warn("Could not open %s as an output file",output_file);
ret = FALSE;
}
rmd = RandomModelDNA_std();
cps = flat_cDNAParser(indel_error);
cm = flat_CodonMapper(ct);
sprinkle_errors_over_CodonMapper(cm,subs_error);
return ret;
}
boolean user_DebugMatrix(DebugMatrix * de)
{
char buffer[MAXLINE];
char ** base;
char ** brk;
FILE * in;
FILE * out;
DebugBreakPoint * bp;
assert(de);
assert(de->in);
assert(de->out);
in = de->in;
out = de->out;
/* set reset to FALSE */
de->reset = 0;
fprintf(out,"Entering dynamite debugger. Type help for help\n");
while(1) {
fprintf(out,"Dy %5d:%5d max: %d >",de->currenti,de->currentj,de->max_score);
fgets(buffer,MAXLINE,in);
if( strstartcmp(buffer,"quit") == 0 ) {
exit(1);
}
if( strstartcmp(buffer,"show") == 0 ) {
show_DebugMatrix(de,buffer);
continue;
}
if( strstartcmp(buffer,"run") == 0 ) {
/* return to calling function */
return 0;
}
if( strstartcmp(buffer,"break") == 0 ) {
base = brk = breakstring(buffer,spacestr);
brk++;
if( *brk == NULL || *(brk+1) == NULL ) {
fprintf(out,">>>> break must have i j positions\n");
continue;
} else {
bp = DebugBreakPoint_alloc();
bp->type = MDBP_Cursor;
/* reset cursor positions */
if( is_integer_string(*brk,&bp->posi) == FALSE ) {
fprintf(out,">>>> i position not an integer.\n");
}
brk++;
if( is_integer_string(*brk,&bp->posj) == FALSE ) {
fprintf(out,">>>> j position not an integer.\n");
}
fprintf(out,"Adding cursor break point %d,%d\n",bp->posi,bp->posj);
add_bp_DebugMatrix(de,bp);
}
ckfree(base);
continue;
}
if( strstartcmp(buffer,"set") == 0 ) {
base = brk = breakstring(buffer,spacestr);
brk++;
if( *brk == NULL || *(brk+1) == NULL ) {
fprintf(out,">>>> set must have i j positions\n");
} else {
/* reset cursor positions */
if( is_integer_string(*brk,&de->currenti) == FALSE ) {
fprintf(out,">>>> i position not an integer.\n");
}
if( is_integer_string(*brk,&de->currenti) == FALSE ) {
fprintf(out,">>>> j position not an integer.\n");
}
}
de->reset = 1;
ckfree(base);
continue;
}
}
}