/* unsafe function that will be replaced by a threadsafe companion in the future */
PRIVATE void init_alifold(int length){
unsigned int n;
if (length<1) nrerror("initialize_fold: argument must be greater 0");
if (init_length>0) free_alifold_arrays();
get_arrays((unsigned) length);
make_pair_matrix();
init_length=length;
for (n = 1; n <= (unsigned) length; n++)
indx[n] = (n*(n-1)) >> 1; /* n(n-1)/2 */
update_fold_params();
}
/* unsafe function that will be replaced by a threadsafe companion in the future */
PRIVATE void init_alifold(int length){
#ifdef _OPENMP
/* Explicitly turn off dynamic threads */
omp_set_dynamic(0);
#endif
if (length < 1) nrerror("initialize_fold: argument must be greater 0");
free_alifold_arrays();
get_arrays((unsigned) length);
init_length = length;
indx = get_indx((unsigned)length);
update_alifold_params();
}
int main(int argc, char *argv[])
{
char *modelDir=NULL; /* Directory with model files */
struct svm_model* decision_model; /* SVM classification model */
/* Command line options */
int reverse=0; /* Scan reverse complement */
int showVersion=0; /* Shows version and exits */
int showHelp=0; /* Show short help and exits */
int from=-1; /* Scan slice from-to */
int to=-1;
FILE *clust_file=stdin; /* Input file */
FILE *out=stdout; /* Output file */
struct aln *AS[MAX_NUM_NAMES];
struct aln *window[MAX_NUM_NAMES];
char *tmpAln[MAX_NUM_NAMES];
int n_seq; /* number of input sequences */
int length; /* length of alignment/window */
int z_score_type;
int decision_model_type;
char *structure=NULL;
char *singleStruc,*gapStruc, *output,*woGapsSeq;
char strand[8];
char warningString[2000];
char warningString_regression[2000];
char *string=NULL;
double singleMFE,sumMFE,singleZ,sumZ,z,sci,id,decValue,prob,comb,entropy,GC;
double min_en, real_en;
int i,j,k,l,ll,r,countAln,nonGaps,singleGC;
int (*readFunction)(FILE *clust,struct aln *alignedSeqs[]);
char** lines=NULL;
int directions[3]={FORWARD,0,0};
int currDirection;
struct gengetopt_args_info args;
double meanMFE_fwd=0;
double consensusMFE_fwd=0;
double sci_fwd=0;
double z_fwd=0;
int strandGuess;
int avoid_shuffle=0;
double strandProb,strandDec;
if (cmdline_parser (argc, argv, &args) != 0){
usage();
exit(EXIT_FAILURE);
}
if (args.help_given){
help();
exit(EXIT_SUCCESS);
}
if (args.version_given){
version();
exit(EXIT_SUCCESS);
}
if (args.outfile_given){
out = fopen(args.outfile_arg, "w");
if (out == NULL){
fprintf(stderr, "ERROR: Can't open output file %s\n", args.outfile_arg);
exit(1);
}
}
/* Strand prediction implies both strands scored */
if (args.predict_strand_flag){
args.both_strands_flag=1;
}
if (args.forward_flag && !args.reverse_flag){
directions[0]=FORWARD;
directions[1]=directions[2]=0;
}
if (!args.forward_flag && args.reverse_flag){
directions[0]=REVERSE;
directions[1]=directions[2]=0;
}
if ((args.forward_flag && args.reverse_flag) || args.both_strands_flag){
directions[0]=FORWARD;
directions[1]=REVERSE;
}
if (args.window_given){
if (sscanf(args.window_arg,"%d-%d",&from,&to)!=2){
nrerror("ERROR: Invalid --window/-w command. "
"Use it like '--window 100-200'\n");
}
printf("from:%d,to:%d\n",from,to);
}
if (args.inputs_num>=1){
clust_file = fopen(args.inputs[0], "r");
if (clust_file == NULL){
fprintf(stderr, "ERROR: Can't open input file %s\n", args.inputs[0]);
exit(1);
}
}
/* Global RNA package variables */
do_backtrack = 1;
dangles=2;
switch(checkFormat(clust_file)){
case CLUSTAL:
readFunction=&read_clustal;
break;
case MAF:
readFunction=&read_maf;
break;
case 0:
nrerror("ERROR: Unknown alignment file format. Use Clustal W or MAF format.\n");
}
/* Set z-score type (mono/dinucleotide) here */
z_score_type = 2;
if (args.mononucleotide_given) z_score_type = 0;
/* now let's decide which decision model to take */
/* decision_model_type = 1 for normal model used in RNAz 1.0 */
/* decision_model_type = 2 for normal model using dinucelotide background */
/* decision_model_type = 3 for structural model using dinucelotide background */
decision_model_type = 2;
if (args.mononucleotide_given) decision_model_type = 1;
if (args.locarnate_given) decision_model_type = 3;
if ((args.mononucleotide_given) && args.locarnate_given){
z_score_type=2;
nrerror("ERROR: Structural decision model only trained with dinucleotide background model.\n");
}
if (args.no_shuffle_given) avoid_shuffle = 1;
decision_model=get_decision_model(NULL, decision_model_type);
/* Initialize Regression Models for mononucleotide */
/* Not needed if we score with dinucleotides */
if (z_score_type == 0) regression_svm_init();
countAln=0;
while ((n_seq=readFunction(clust_file, AS))!=0){
if (n_seq ==1){
nrerror("ERROR: You need at least two sequences in the alignment.\n");
}
countAln++;
length = (int) strlen(AS[0]->seq);
/* if a slice is specified by the user */
if ((from!=-1 || to!=-1) && (countAln==1)){
if ((from>=to)||(from<=0)||(to>length)){
nrerror("ERROR: Invalid window range given.\n");
}
sliceAln((const struct aln**)AS, (struct aln **)window, from, to);
length=to-from+1;
} else { /* take complete alignment */
/* window=AS does not work..., deep copy seems not necessary here*/
from=1;
to=length;
sliceAln((const struct aln **)AS, (struct aln **)window, 1, length);
}
/* Convert all Us to Ts for RNAalifold. There is a slight
difference in the results. During training we used alignments
with Ts, so we use Ts here as well. */
for (i=0;i<n_seq;i++){
j=0;
while (window[i]->seq[j]){
window[i]->seq[j]=toupper(window[i]->seq[j]);
if (window[i]->seq[j]=='U') window[i]->seq[j]='T';
++j;
}
}
k=0;
while ((currDirection=directions[k++])!=0){
if (currDirection==REVERSE){
revAln((struct aln **)window);
strcpy(strand,"reverse");
} else {
strcpy(strand,"forward");
}
structure = (char *) space((unsigned) length+1);
for (i=0;window[i]!=NULL;i++){
tmpAln[i]=window[i]->seq;
}
tmpAln[i]=NULL;
min_en = alifold(tmpAln, structure);
free_alifold_arrays();
comb=combPerPair(window,structure);
sumZ=0.0;
sumMFE=0.0;
GC=0.0;
output=(char *)space(sizeof(char)*(length+160)*(n_seq+1)*3);
strcpy(warningString,"");
strcpy(warningString_regression,"");
for (i=0;i<n_seq;i++){
singleStruc = space(strlen(window[i]->seq)+1);
woGapsSeq = space(strlen(window[i]->seq)+1);
j=0;
nonGaps=0;
singleGC=0;
while (window[i]->seq[j]){
/* Convert all Ts to Us for RNAfold. There is a difference
between the results. With U in the function call, we get
the results as RNAfold gives on the command line. Since
this variant was also used during training, we use it here
as well. */
if (window[i]->seq[j]=='T') window[i]->seq[j]='U';
if (window[i]->seq[j]=='C') singleGC++;
if (window[i]->seq[j]=='G') singleGC++;
if (window[i]->seq[j]!='-'){
nonGaps++;
woGapsSeq[strlen(woGapsSeq)]=window[i]->seq[j];
woGapsSeq[strlen(woGapsSeq)]='\0';
}
++j;
}
/* z-score is calculated here! */
singleMFE = fold(woGapsSeq, singleStruc);
free_arrays();
/* z-score type may be overwritten. If it is out of training
bounds, we switch to shuffling if allowed (avoid_shuffle). */
int z_score_type_orig = z_score_type;
singleZ=mfe_zscore(woGapsSeq,singleMFE, &z_score_type, avoid_shuffle, warningString_regression);
GC+=(double) singleGC/nonGaps;
sumZ+=singleZ;
sumMFE+=singleMFE;
if (window[1]->strand!='?' && !args.window_given){
sprintf(output+strlen(output),
">%s %d %d %c %d\n",
window[i]->name,window[i]->start,
window[i]->length,window[i]->strand,
window[i]->fullLength);
} else {
sprintf(output+strlen(output),">%s\n",window[i]->name);
}
gapStruc= (char *) space(sizeof(char)*(strlen(window[i]->seq)+1));
l=ll=0;
while (window[i]->seq[l]!='\0'){
if (window[i]->seq[l]!='-'){
gapStruc[l]=singleStruc[ll];
l++;
ll++;
} else {
gapStruc[l]='-';
l++;
}
}
char ch;
ch = 'R';
if (z_score_type == 1 || z_score_type == 3) ch = 'S';
sprintf(output+strlen(output),"%s\n%s ( %6.2f, z-score = %6.2f, %c)\n",
window[i]->seq,gapStruc,singleMFE,singleZ,ch);
z_score_type = z_score_type_orig;
free(woGapsSeq);
free(singleStruc);
}
{
int i; double s=0;
extern int eos_debug;
eos_debug=-1; /* shut off warnings about nonstandard pairs */
for (i=0; window[i]!=NULL; i++)
s += energy_of_struct(window[i]->seq, structure);
real_en = s/i;
}
string = consensus((const struct aln**) window);
sprintf(output+strlen(output),
">consensus\n%s\n%s (%6.2f = %6.2f + %6.2f) \n",
string, structure, min_en, real_en, min_en-real_en );
free(string);
id=meanPairID((const struct aln**)window);
entropy=NormShannonEntropy((const struct aln**)window);
z=sumZ/n_seq;
GC=(double)GC/n_seq;
if (sumMFE==0){
/*Set SCI to 0 in the weird case of no structure in single
sequences*/
sci=0;
} else {
sci=min_en/(sumMFE/n_seq);
}
decValue=999;
prob=0;
classify(&prob,&decValue,decision_model,id,n_seq,z,sci,entropy,decision_model_type);
if (args.cutoff_given){
if (prob<args.cutoff_arg){
continue;
}
}
warning(warningString,id,n_seq,z,sci,entropy,(struct aln **)window,decision_model_type);
fprintf(out,"\n############################ RNAz "PACKAGE_VERSION" ##############################\n\n");
fprintf(out," Sequences: %u\n", n_seq);
if (args.window_given){
fprintf(out," Slice: %u to %u\n",from,to);
}
fprintf(out," Columns: %u\n",length);
fprintf(out," Reading direction: %s\n",strand);
fprintf(out," Mean pairwise identity: %6.2f\n", id);
fprintf(out," Shannon entropy: %2.5f\n", entropy);
fprintf(out," G+C content: %2.5f\n", GC);
fprintf(out," Mean single sequence MFE: %6.2f\n", sumMFE/n_seq);
fprintf(out," Consensus MFE: %6.2f\n",min_en);
fprintf(out," Energy contribution: %6.2f\n",real_en);
fprintf(out," Covariance contribution: %6.2f\n",min_en-real_en);
fprintf(out," Combinations/Pair: %6.2f\n",comb);
fprintf(out," Mean z-score: %6.2f\n",z);
fprintf(out," Structure conservation index: %6.2f\n",sci);
if (decision_model_type == 1) {
fprintf(out," Background model: mononucleotide\n");
fprintf(out," Decision model: sequence based alignment quality\n");
}
if (decision_model_type == 2) {
fprintf(out," Background model: dinucleotide\n");
fprintf(out," Decision model: sequence based alignment quality\n");
}
if (decision_model_type == 3) {
fprintf(out," Background model: dinucleotide\n");
fprintf(out," Decision model: structural RNA alignment quality\n");
}
fprintf(out," SVM decision value: %6.2f\n",decValue);
fprintf(out," SVM RNA-class probability: %6f\n",prob);
if (prob>0.5){
fprintf(out," Prediction: RNA\n");
}
else {
fprintf(out," Prediction: OTHER\n");
}
fprintf(out,"%s",warningString_regression);
fprintf(out,"%s",warningString);
fprintf(out,"\n######################################################################\n\n");
fprintf(out,"%s",output);
fflush(out);
free(structure);
free(output);
if (currDirection==FORWARD && args.predict_strand_flag){
meanMFE_fwd=sumMFE/n_seq;
consensusMFE_fwd=min_en;
sci_fwd=sci;
z_fwd=z;
}
if (currDirection==REVERSE && args.predict_strand_flag){
if (predict_strand(sci_fwd-sci, meanMFE_fwd-(sumMFE/n_seq),
consensusMFE_fwd-min_en, z_fwd-z, n_seq, id,
&strandGuess, &strandProb, &strandDec, NULL)){
if (strandGuess==1){
fprintf(out, "\n# Strand winner: forward (%.2f)\n",strandProb);
} else {
fprintf(out, "\n# Strand winner: reverse (%.2f)\n",1-strandProb);
}
} else {
fprintf(out, "\n# WARNING: No strand prediction (values out of range)\n");
}
}
}
freeAln((struct aln **)AS);
freeAln((struct aln **)window);
}
if (args.inputs_num>=1){
fclose(clust_file);
}
cmdline_parser_free (&args);
if (countAln==0){
nrerror("ERROR: Empty alignment file\n");
}
svm_destroy_model(decision_model);
regression_svm_free();
return 0;
}
/*--------------------------------------------------------------------------*/
int main(int argc, char *argv[]){
struct RNAalifold_args_info args_info;
unsigned int input_type;
char ffname[FILENAME_MAX_LENGTH], gfname[FILENAME_MAX_LENGTH], fname[FILENAME_MAX_LENGTH];
char *input_string, *string, *structure, *cstruc, *ParamFile, *ns_bases, *c;
int n_seq, i, length, sym, r, noPS, with_sci;
int endgaps, mis, circular, doAlnPS, doColor, doMEA, n_back, eval_energy, pf, istty;
double min_en, real_en, sfact, MEAgamma, bppmThreshold, betaScale;
char *AS[MAX_NUM_NAMES]; /* aligned sequences */
char *names[MAX_NUM_NAMES]; /* sequence names */
FILE *clust_file = stdin;
pf_paramT *pf_parameters;
model_detailsT md;
fname[0] = ffname[0] = gfname[0] = '\0';
string = structure = cstruc = ParamFile = ns_bases = NULL;
pf_parameters = NULL;
endgaps = mis = pf = circular = doAlnPS = doColor = n_back = eval_energy = oldAliEn = doMEA = ribo = noPS = 0;
do_backtrack = 1;
dangles = 2;
gquad = 0;
sfact = 1.07;
bppmThreshold = 1e-6;
MEAgamma = 1.0;
betaScale = 1.;
with_sci = 0;
set_model_details(&md);
/*
#############################################
# check the command line prameters
#############################################
*/
if(RNAalifold_cmdline_parser (argc, argv, &args_info) != 0) exit(1);
/* temperature */
if(args_info.temp_given) temperature = args_info.temp_arg;
/* structure constraint */
if(args_info.constraint_given) fold_constrained=1;
/* do not take special tetra loop energies into account */
if(args_info.noTetra_given) md.special_hp = tetra_loop=0;
/* set dangle model */
if(args_info.dangles_given){
if((args_info.dangles_arg != 0) && (args_info.dangles_arg != 2))
warn_user("required dangle model not implemented, falling back to default dangles=2");
else
md.dangles = dangles=args_info.dangles_arg;
}
/* do not allow weak pairs */
if(args_info.noLP_given) md.noLP = noLonelyPairs = 1;
/* do not allow wobble pairs (GU) */
if(args_info.noGU_given) md.noGU = noGU = 1;
/* do not allow weak closing pairs (AU,GU) */
if(args_info.noClosingGU_given) md.noGUclosure = no_closingGU = 1;
/* gquadruplex support */
if(args_info.gquad_given) md.gquad = gquad = 1;
/* sci computation */
if(args_info.sci_given) with_sci = 1;
/* do not convert DNA nucleotide "T" to appropriate RNA "U" */
/* set energy model */
if(args_info.energyModel_given) energy_set = args_info.energyModel_arg;
/* take another energy parameter set */
if(args_info.paramFile_given) ParamFile = strdup(args_info.paramFile_arg);
/* Allow other pairs in addition to the usual AU,GC,and GU pairs */
if(args_info.nsp_given) ns_bases = strdup(args_info.nsp_arg);
/* set pf scaling factor */
if(args_info.pfScale_given) sfact = args_info.pfScale_arg;
/* assume RNA sequence to be circular */
if(args_info.circ_given) circular=1;
/* do not produce postscript output */
if(args_info.noPS_given) noPS = 1;
/* partition function settings */
if(args_info.partfunc_given){
pf = 1;
if(args_info.partfunc_arg != -1)
do_backtrack = args_info.partfunc_arg;
}
/* MEA (maximum expected accuracy) settings */
if(args_info.MEA_given){
pf = doMEA = 1;
if(args_info.MEA_arg != -1)
MEAgamma = args_info.MEA_arg;
}
if(args_info.betaScale_given) betaScale = args_info.betaScale_arg;
/* set the bppm threshold for the dotplot */
if(args_info.bppmThreshold_given)
bppmThreshold = MIN2(1., MAX2(0.,args_info.bppmThreshold_arg));
/* set cfactor */
if(args_info.cfactor_given) cv_fact = args_info.cfactor_arg;
/* set nfactor */
if(args_info.nfactor_given) nc_fact = args_info.nfactor_arg;
if(args_info.endgaps_given) endgaps = 1;
if(args_info.mis_given) mis = 1;
if(args_info.color_given) doColor=1;
if(args_info.aln_given) doAlnPS=1;
if(args_info.old_given) oldAliEn = 1;
if(args_info.stochBT_given){
n_back = args_info.stochBT_arg;
do_backtrack = 0;
pf = 1;
init_rand();
}
if(args_info.stochBT_en_given){
n_back = args_info.stochBT_en_arg;
do_backtrack = 0;
pf = 1;
eval_energy = 1;
init_rand();
}
if(args_info.ribosum_file_given){
RibosumFile = strdup(args_info.ribosum_file_arg);
ribo = 1;
}
if(args_info.ribosum_scoring_given){
RibosumFile = NULL;
ribo = 1;
}
if(args_info.layout_type_given)
rna_plot_type = args_info.layout_type_arg;
/* alignment file name given as unnamed option? */
if(args_info.inputs_num == 1){
clust_file = fopen(args_info.inputs[0], "r");
if (clust_file == NULL) {
fprintf(stderr, "can't open %s\n", args_info.inputs[0]);
}
}
/* free allocated memory of command line data structure */
RNAalifold_cmdline_parser_free (&args_info);
/*
#############################################
# begin initializing
#############################################
*/
if(circular && gquad){
nrerror("G-Quadruplex support is currently not available for circular RNA structures");
}
make_pair_matrix();
if (circular && noLonelyPairs)
warn_user("depending on the origin of the circular sequence, "
"some structures may be missed when using --noLP\n"
"Try rotating your sequence a few times\n");
if (ParamFile != NULL) read_parameter_file(ParamFile);
if (ns_bases != NULL) {
nonstandards = space(33);
c=ns_bases;
i=sym=0;
if (*c=='-') {
sym=1; c++;
}
while (*c!='\0') {
if (*c!=',') {
nonstandards[i++]=*c++;
nonstandards[i++]=*c;
if ((sym)&&(*c!=*(c-1))) {
nonstandards[i++]=*c;
nonstandards[i++]=*(c-1);
}
}
c++;
}
}
istty = isatty(fileno(stdout))&&isatty(fileno(stdin));
/*
########################################################
# handle user input from 'stdin' if necessary
########################################################
*/
if(fold_constrained){
if(istty){
print_tty_constraint_full();
print_tty_input_seq_str("");
}
input_type = get_input_line(&input_string, VRNA_INPUT_NOSKIP_COMMENTS);
if(input_type & VRNA_INPUT_QUIT){ return 0;}
else if((input_type & VRNA_INPUT_MISC) && (strlen(input_string) > 0)){
cstruc = strdup(input_string);
free(input_string);
}
else warn_user("constraints missing");
}
if (istty && (clust_file == stdin))
print_tty_input_seq_str("Input aligned sequences in clustalw or stockholm format\n(enter a line starting with \"//\" to indicate the end of your input)");
n_seq = read_clustal(clust_file, AS, names);
if (n_seq==0) nrerror("no sequences found");
if (clust_file != stdin) fclose(clust_file);
/*
########################################################
# done with 'stdin' handling, now init everything properly
########################################################
*/
length = (int) strlen(AS[0]);
structure = (char *)space((unsigned) length+1);
if(fold_constrained && cstruc != NULL)
strncpy(structure, cstruc, length);
if (endgaps)
for (i=0; i<n_seq; i++) mark_endgaps(AS[i], '~');
/*
########################################################
# begin actual calculations
########################################################
*/
if (circular) {
int i;
double s = 0;
min_en = circalifold((const char **)AS, structure);
for (i=0; AS[i]!=NULL; i++)
s += energy_of_circ_structure(AS[i], structure, -1);
real_en = s/i;
} else {
float *ens = (float *)space(2*sizeof(float));
min_en = alifold((const char **)AS, structure);
if(md.gquad)
energy_of_ali_gquad_structure((const char **)AS, structure, n_seq, ens);
else
energy_of_alistruct((const char **)AS, structure, n_seq, ens);
real_en = ens[0];
free(ens);
}
string = (mis) ? consens_mis((const char **) AS) : consensus((const char **) AS);
printf("%s\n%s", string, structure);
if(istty){
if(with_sci){
float sci = min_en;
float e_mean = 0;
for (i=0; AS[i]!=NULL; i++){
char *seq = get_ungapped_sequence(AS[i]);
char *str = (char *)space(sizeof(char) * (strlen(seq) + 1));
e_mean += fold(seq, str);
free(seq);
free(str);
}
e_mean /= i;
sci /= e_mean;
printf( "\n minimum free energy = %6.2f kcal/mol (%6.2f + %6.2f)"
"\n SCI = %2.4f\n",
min_en, real_en, min_en-real_en, sci);
} else
printf("\n minimum free energy = %6.2f kcal/mol (%6.2f + %6.2f)\n",
min_en, real_en, min_en - real_en);
} else {
if(with_sci){
float sci = min_en;
float e_mean = 0;
for (i=0; AS[i]!=NULL; i++){
char *seq = get_ungapped_sequence(AS[i]);
char *str = (char *)space(sizeof(char) * (strlen(seq) + 1));
e_mean += fold(seq, str);
free(seq);
free(str);
}
e_mean /= i;
sci /= e_mean;
printf(" (%6.2f = %6.2f + %6.2f) [%2.4f]\n", min_en, real_en, min_en-real_en, sci);
}
else
printf(" (%6.2f = %6.2f + %6.2f) \n", min_en, real_en, min_en-real_en );
}
strcpy(ffname, "alirna.ps");
strcpy(gfname, "alirna.g");
if (!noPS) {
char **A;
A = annote(structure, (const char**) AS);
if(md.gquad){
if (doColor)
(void) PS_rna_plot_a_gquad(string, structure, ffname, A[0], A[1]);
else
(void) PS_rna_plot_a_gquad(string, structure, ffname, NULL, A[1]);
} else {
if (doColor)
(void) PS_rna_plot_a(string, structure, ffname, A[0], A[1]);
else
(void) PS_rna_plot_a(string, structure, ffname, NULL, A[1]);
}
free(A[0]); free(A[1]); free(A);
}
if (doAlnPS)
PS_color_aln(structure, "aln.ps", (const char const **) AS, (const char const **) names);
/* free mfe arrays */
free_alifold_arrays();
if (pf) {
float energy, kT;
char * mfe_struc;
mfe_struc = strdup(structure);
kT = (betaScale*((temperature+K0)*GASCONST))/1000.; /* in Kcal */
pf_scale = exp(-(sfact*min_en)/kT/length);
if (length>2000) fprintf(stderr, "scaling factor %f\n", pf_scale);
fflush(stdout);
if (cstruc!=NULL) strncpy(structure, cstruc, length+1);
pf_parameters = get_boltzmann_factors_ali(n_seq, temperature, betaScale, md, pf_scale);
energy = alipf_fold_par((const char **)AS, structure, NULL, pf_parameters, do_backtrack, fold_constrained, circular);
if (n_back>0) {
/*stochastic sampling*/
for (i=0; i<n_back; i++) {
char *s;
double prob=1.;
s = alipbacktrack(&prob);
printf("%s ", s);
if (eval_energy ) printf("%6g %.2f ",prob, -1*(kT*log(prob)-energy));
printf("\n");
free(s);
}
}
if (do_backtrack) {
printf("%s", structure);
if (!istty) printf(" [%6.2f]\n", energy);
else printf("\n");
}
if ((istty)||(!do_backtrack))
printf(" free energy of ensemble = %6.2f kcal/mol\n", energy);
printf(" frequency of mfe structure in ensemble %g\n",
exp((energy-min_en)/kT));
if (do_backtrack) {
FILE *aliout;
cpair *cp;
char *cent;
double dist;
FLT_OR_DBL *probs = export_ali_bppm();
plist *pl, *mfel;
assign_plist_from_pr(&pl, probs, length, bppmThreshold);
assign_plist_from_db(&mfel, mfe_struc, 0.95*0.95);
if (!circular){
float *ens;
cent = get_centroid_struct_pr(length, &dist, probs);
ens=(float *)space(2*sizeof(float));
energy_of_alistruct((const char **)AS, cent, n_seq, ens);
/*cent_en = energy_of_struct(string, cent);*/ /*ali*/
printf("%s %6.2f {%6.2f + %6.2f}\n",cent,ens[0]-ens[1],ens[0],(-1)*ens[1]);
free(cent);
free(ens);
}
if(doMEA){
float mea, *ens;
plist *pl2;
assign_plist_from_pr(&pl2, probs, length, 1e-4/(1+MEAgamma));
mea = MEA(pl2, structure, MEAgamma);
ens = (float *)space(2*sizeof(float));
if(circular)
energy_of_alistruct((const char **)AS, structure, n_seq, ens);
else
ens[0] = energy_of_structure(string, structure, 0);
printf("%s {%6.2f MEA=%.2f}\n", structure, ens[0], mea);
free(ens);
free(pl2);
}
if (fname[0]!='\0') {
strcpy(ffname, fname);
strcat(ffname, "_ali.out");
} else strcpy(ffname, "alifold.out");
aliout = fopen(ffname, "w");
if (!aliout) {
fprintf(stderr, "can't open %s skipping output\n", ffname);
} else {
print_aliout(AS, pl, bppmThreshold, n_seq, mfe_struc, aliout);
}
fclose(aliout);
if (fname[0]!='\0') {
strcpy(ffname, fname);
strcat(ffname, "_dp.ps");
} else strcpy(ffname, "alidot.ps");
cp = make_color_pinfo(AS,pl, bppmThreshold, n_seq, mfel);
(void) PS_color_dot_plot(string, cp, ffname);
free(cp);
free(pl);
free(mfel);
}
free(mfe_struc);
free_alipf_arrays();
free(pf_parameters);
}
if (cstruc!=NULL) free(cstruc);
(void) fflush(stdout);
free(string);
free(structure);
for (i=0; AS[i]; i++) {
free(AS[i]); free(names[i]);
}
return 0;
}
int main(int argc, char *argv[])
{
char *string;
char *structure=NULL;
char *cstruc=NULL;
char *ns_bases=NULL;
char *c;
int n_seq;
int i;
int length;
int sym;
int endgaps = 0;
int mis = 0;
double min_en;
double real_en;
double sfact = 1.07;
int pf = 0;
int istty;
char *AS[MAX_NUM_NAMES]; /* aligned sequences */
char *names[MAX_NUM_NAMES]; /* sequence names */
AjPSeqset seq = NULL;
AjPFile confile = NULL;
AjPFile alifile = NULL;
AjPFile paramfile = NULL;
AjPFile outf = NULL;
AjPFile essfile = NULL;
AjPFile dotfile = NULL;
AjPStr constring = NULL;
float eT = 0.;
AjBool eGU;
AjBool eclose;
AjBool lonely;
AjPStr ensbases = NULL;
AjBool etloop;
AjPStr eenergy = NULL;
char ewt = '\0';
float escale = 0.;
AjPStr edangles = NULL;
char edangle = '\0';
ajint len;
AjPSeq tseq = NULL;
AjPStr tname = NULL;
int circ = 0;
int doAlnPS = 0;
int doColor = 0;
embInitPV("vrnaalifoldpf",argc,argv,"VIENNA",VERSION);
constring = ajStrNew();
seq = ajAcdGetSeqset("sequence");
confile = ajAcdGetInfile("constraintfile");
paramfile = ajAcdGetInfile("paramfile");
eT = ajAcdGetFloat("temperature");
eGU = ajAcdGetBoolean("gu");
eclose = ajAcdGetBoolean("closegu");
lonely = ajAcdGetBoolean("lp");
ensbases = ajAcdGetString("nsbases");
etloop = ajAcdGetBoolean("tetraloop");
eenergy = ajAcdGetListSingle("energy");
escale = ajAcdGetFloat("scale");
edangles = ajAcdGetListSingle("dangles");
mis = !!ajAcdGetBoolean("most");
endgaps = !!ajAcdGetBoolean("endgaps");
nc_fact = (double) ajAcdGetFloat("nspenalty");
cv_fact = (double) ajAcdGetFloat("covariance");
outf = ajAcdGetOutfile("outfile");
essfile = ajAcdGetOutfile("ssoutfile");
alifile = ajAcdGetOutfile("alignoutfile");
circ = !!ajAcdGetBoolean("circular");
doColor = !!ajAcdGetBoolean("colour");
dotfile = ajAcdGetOutfile("dotoutfile");
do_backtrack = 1;
pf = 1;
string = NULL;
istty = 0;
dangles = 2;
temperature = (double) eT;
noGU = (eGU) ? 0 : 1;
no_closingGU = (eclose) ? 0 : 1;
noLonelyPairs = (lonely) ? 0 : 1;
ns_bases = (ajStrGetLen(ensbases)) ? MAJSTRGETPTR(ensbases) : NULL;
tetra_loop = !!etloop;
ewt = *ajStrGetPtr(eenergy);
if(ewt == '0')
energy_set = 0;
else if(ewt == '1')
energy_set = 1;
else if(ewt == '2')
energy_set = 2;
sfact = (double) escale;
edangle = *ajStrGetPtr(edangles);
if(edangle == '0')
dangles = 0;
else if(edangle == '1')
dangles = 1;
else if(edangle == '2')
dangles = 2;
else if(edangle == '3')
dangles = 3;
if(paramfile)
read_parameter_file(paramfile);
if (ns_bases != NULL)
{
nonstandards = space(33);
c=ns_bases;
i=sym=0;
if (*c=='-')
{
sym=1;
c++;
}
while (*c!='\0')
{
if (*c!=',')
{
nonstandards[i++]=*c++;
nonstandards[i++]=*c;
if ((sym)&&(*c!=*(c-1)))
{
nonstandards[i++]=*c;
nonstandards[i++]=*(c-1);
}
}
c++;
}
}
if(alifile)
doAlnPS = 1;
if(confile)
vienna_GetConstraints(confile,&constring);
n_seq = ajSeqsetGetSize(seq);
if(n_seq > MAX_NUM_NAMES - 1)
ajFatal("[e]RNAalifold is restricted to %d sequences\n",
MAX_NUM_NAMES - 1);
if (n_seq==0)
ajFatal("No sequences found");
for(i=0;i<n_seq;++i)
{
tseq = (AjPSeq) ajSeqsetGetseqSeq(seq,i);
ajSeqGapStandard(tseq, '-');
tname = (AjPStr) ajSeqsetGetseqNameS(seq,i);
len = ajSeqGetLen(tseq);
AS[i] = (char *) space(len+1);
names[i] = (char *) space(ajStrGetLen(tname)+1);
strcpy(AS[i],ajSeqGetSeqC(tseq));
strcpy(names[i],ajStrGetPtr(tname));
}
AS[n_seq] = NULL;
names[n_seq] = NULL;
if (endgaps)
for (i=0; i<n_seq; i++)
mark_endgaps(AS[i], '~');
length = (int) strlen(AS[0]);
structure = (char *) space((unsigned) length+1);
if(confile)
{
fold_constrained = 1;
strcpy(structure,ajStrGetPtr(constring));
}
if (circ && noLonelyPairs)
ajWarn(
"warning, depending on the origin of the circular sequence, "
"some structures may be missed when using -noLP\n"
"Try rotating your sequence a few times\n");
if (circ)
min_en = circalifold((const char **)AS, structure);
else
min_en = alifold(AS, structure);
{
int i;
double s=0;
extern int eos_debug;
eos_debug=-1; /* shut off warnings about nonstandard pairs */
for (i=0; AS[i]!=NULL; i++)
if (circ)
s += energy_of_circ_struct(AS[i], structure);
else
s += energy_of_struct(AS[i], structure);
real_en = s/i;
}
string = (mis) ?
consens_mis((const char **) AS) : consensus((const char **) AS);
ajFmtPrintF(outf,"%s\n%s", string, structure);
ajFmtPrintF(outf," (%6.2f = %6.2f + %6.2f) \n", min_en, real_en,
min_en-real_en );
if (length<=2500) {
char **A;
A = annote(structure, (const char**) AS);
if (doColor)
(void) PS_rna_plot_a(string, structure, essfile, A[0], A[1]);
else
(void) PS_rna_plot_a(string, structure, essfile, NULL, A[1]);
free(A[0]); free(A[1]);free(A);
} else
ajWarn("INFO: structure too long, not doing xy_plot\n");
if (doAlnPS)
PS_color_aln(structure, alifile, AS, names);
{ /* free mfe arrays but preserve base_pair for PS_dot_plot */
struct bond *bp;
bp = base_pair; base_pair = space(16);
free_alifold_arrays(); /* free's base_pair */
free_alipf_arrays();
base_pair = bp;
}
if (pf) {
double energy, kT;
pair_info *pi;
char * mfe_struc;
mfe_struc = strdup(structure);
kT = (temperature+273.15)*1.98717/1000.; /* in Kcal */
pf_scale = exp(-(sfact*min_en)/kT/length);
if (length>2000)
ajWarn("scaling factor %f\n", pf_scale);
/* init_alipf_fold(length); */
if (confile)
strncpy(structure, ajStrGetPtr(constring), length+1);
energy = (circ) ? alipf_circ_fold(AS, structure, &pi) : alipf_fold(AS, structure, &pi);
if (do_backtrack) {
ajFmtPrintF(outf,"%s", structure);
ajFmtPrintF(outf," [%6.2f]\n", energy);
}
if ((istty)||(!do_backtrack))
ajFmtPrintF(outf," free energy of ensemble = %6.2f kcal/mol\n", energy);
ajFmtPrintF(outf," frequency of mfe structure in ensemble %g\n",
exp((energy-min_en)/kT));
if (do_backtrack) {
FILE *aliout;
cpair *cp;
short *ptable; int k;
ptable = make_pair_table(mfe_struc);
ajFmtPrintF(outf,"\n# Alignment section\n\n");
aliout = ajFileGetFileptr(outf);
fprintf(aliout, "%d sequences; length of alignment %d\n",
n_seq, length);
fprintf(aliout, "alifold output\n");
for (k=0; pi[k].i>0; k++) {
pi[k].comp = (ptable[pi[k].i] == pi[k].j) ? 1:0;
print_pi(pi[k], aliout);
}
fprintf(aliout, "%s\n", structure);
free(ptable);
cp = make_color_pinfo(pi);
(void) PS_color_dot_plot(string, cp, dotfile);
free(cp);
free(mfe_struc);
free(pi);
}
}
if (cstruc!=NULL) free(cstruc);
free(base_pair);
(void) fflush(stdout);
free(string);
free(structure);
for (i=0; AS[i]; i++) {
free(AS[i]); free(names[i]);
}
ajSeqsetDel(&seq);
ajStrDel(&constring);
ajStrDel(&eenergy);
ajStrDel(&edangles);
ajStrDel(&ensbases);
ajFileClose(&confile);
ajFileClose(¶mfile);
ajFileClose(&outf);
ajFileClose(&essfile);
ajFileClose(&alifile);
ajFileClose(&dotfile);
embExit();
return 0;
}
