// Read input file (HMM, HHM, or alignment format), and add pseudocounts etc.
void ReadAndPrepare(char* infile, HMM& q, Alignment* qali=NULL)
{
char path[NAMELEN];
// Open query file and determine file type
char line[LINELEN]=""; // input line
FILE* inf=NULL;
if (strcmp(infile,"stdin"))
{
inf = fopen(infile, "r");
if (!inf) OpenFileError(infile);
Pathname(path,infile);
}
else
{
inf = stdin;
if (v>=2) printf("Reading HMM / multiple alignment from standard input ...\n(To get a help list instead, quit and type %s -h.)\n",program_name);
*path='\0';
}
fgetline(line,LINELEN-1,inf);
// Is it an hhm file?
if (!strncmp(line,"NAME",4) || !strncmp(line,"HH",2))
{
if (v>=2) cout<<"Query file is in HHM format\n";
// Rewind to beginning of line and read query hhm file
rewind(inf);
q.Read(inf,path);
if (v>=2 && q.Neff_HMM>11.0)
fprintf(stderr,"WARNING: HMM %s looks too diverse (Neff=%.1f>11). Better check the underlying alignment... \n",q.name,q.Neff_HMM);
// Add transition pseudocounts to query -> q.p[i][a]
q.AddTransitionPseudocounts();
if (!*par.clusterfile) { //compute context-specific pseudocounts?
// Generate an amino acid frequency matrix from f[i][a] with full pseudocount admixture (tau=1) -> g[i][a]
q.PreparePseudocounts();
// Add amino acid pseudocounts to query: q.p[i][a] = (1-tau)*f[i][a] + tau*g[i][a]
q.AddAminoAcidPseudocounts(par.pcm, par.pca, par.pcb, par.pcc);;
} else {
// Add context specific pseudocount to query
q.AddContextSpecificPseudocounts(par.pcm);
}
q.CalculateAminoAcidBackground();
}
// ... or is it an a2m/a3m alignment file
else if (line[0]=='#' || line[0]=='>')
{
Alignment* pali;
if (qali==NULL) pali=new(Alignment); else pali=qali;
if (par.calibrate) {
printf("\nError in %s: only HHM files can be calibrated.\n",program_name);
printf("Build an HHM file from your alignment with 'hhmake -i %s' and rerun hhsearch with the hhm file\n\n",infile);
exit(1);
}
if (v>=2 && strcmp(infile,"stdin")) cout<<infile<<" is in A2M, A3M or FASTA format\n";
// Read alignment from infile into matrix X[k][l] as ASCII (and supply first line as extra argument)
pali->Read(inf,infile,line);
// Convert ASCII to int (0-20),throw out all insert states, record their number in I[k][i]
// and store marked sequences in name[k] and seq[k]
pali->Compress(infile);
// Sort out the nseqdis most dissimilar sequences for display in the output alignments
pali->FilterForDisplay(par.max_seqid,par.coverage,par.qid,par.qsc,par.nseqdis);
// Remove sequences with seq. identity larger than seqid percent (remove the shorter of two)
pali->N_filtered = pali->Filter(par.max_seqid,par.coverage,par.qid,par.qsc,par.Ndiff);
if (par.Neff>=0.999)
pali->FilterNeff();
// Calculate pos-specific weights, AA frequencies and transitions -> f[i][a], tr[i][a]
pali->FrequenciesAndTransitions(q);
if (v>=2 && q.Neff_HMM>11.0)
fprintf(stderr,"WARNING: alignment %s looks too diverse (Neff=%.1f>11). Better check it with an alignment viewer... \n",q.name,q.Neff_HMM);
// Add transition pseudocounts to query -> p[i][a]
q.AddTransitionPseudocounts();
if (!*par.clusterfile) { //compute context-specific pseudocounts?
// Generate an amino acid frequency matrix from f[i][a] with full pseudocount admixture (tau=1) -> g[i][a]
q.PreparePseudocounts();
// Add amino acid pseudocounts to query: p[i][a] = (1-tau)*f[i][a] + tau*g[i][a]
q.AddAminoAcidPseudocounts(par.pcm, par.pca, par.pcb, par.pcc);
} else {
// Add context specific pseudocount to query
q.AddContextSpecificPseudocounts(par.pcm);
}
q.CalculateAminoAcidBackground();
if (qali==NULL) delete(pali);
} else if (!strncmp(line,"HMMER",5)) {
///////////////////////////////////////////////////////////////////////////////////////
// Don't allow HMMER format as input due to the severe loss of sensitivity!!!! (only allowed in HHmake)
if (strncmp(program_name,"hhmake",6)) {
cerr<<endl<<"Error in "<<program_name<<": HMMER format not allowed as input due to the severe loss of sensitivity!\n";
exit(1);
}
// Is infile a HMMER3 file?
if (!strncmp(line,"HMMER3",6))
{
if (v>=2) cout<<"Query file is in HMMER3 format\n";
// Read 'query HMMER file
rewind(inf);
q.ReadHMMer3(inf,path);
// Don't add transition pseudocounts to query!!
// DON'T ADD amino acid pseudocounts to query: pcm=0! q.p[i][a] = f[i][a]
q.AddAminoAcidPseudocounts(0, par.pca, par.pcb, par.pcc);
q.CalculateAminoAcidBackground();
}
// ... or is infile an old HMMER file?
else if (!strncmp(line,"HMMER",5))
{
if (v>=2) cout<<"Query file is in HMMER format\n";
// Read 'query HMMER file
rewind(inf);
q.ReadHMMer(inf,path);
// DON'T ADD amino acid pseudocounts to query: pcm=0! q.p[i][a] = f[i][a]
q.AddAminoAcidPseudocounts(0, par.pca, par.pcb, par.pcc);
q.CalculateAminoAcidBackground();
}
} else {
cerr<<endl<<"Error in "<<program_name<<": unrecognized input file format in \'"<<infile<<"\'\n";
cerr<<"line = "<<line<<"\n";
exit(1);
}
fclose(inf);
if (par.addss==1)
CalculateSS(q);
if (par.columnscore == 5 && !q.divided_by_local_bg_freqs) q.DivideBySqrtOfLocalBackgroundFreqs(par.half_window_size_local_aa_bg_freqs);
if (par.forward>=1) q.Log2LinTransitionProbs(1.0);
return;
}
// Read input file (HMM, HHM, or alignment format), and add pseudocounts etc.
void ReadInput(char* infile, HMM& q, Alignment* qali=NULL)
{
char path[NAMELEN];
// Open query file and determine file type
char line[LINELEN]=""; // input line
FILE* inf=NULL;
if (strcmp(infile,"stdin"))
{
inf = fopen(infile, "r");
if (!inf) OpenFileError(infile);
Pathname(path,infile);
}
else
{
inf = stdin;
if (v>=2) printf("Reading HMM / multiple alignment from standard input ...\n(To get a help list instead, quit and type %s -h.)\n",program_name);
*path='\0';
}
fgetline(line,LINELEN-1,inf);
// Is infile a HMMER3 file?
if (!strncmp(line,"HMMER3",6))
{
if (v>=2) cout<<"Query file is in HMMER3 format\n";
cerr<<"WARNING: Use of HMMER3 format as input will result in severe loss of sensitivity!\n";
// Read 'query HMMER file
rewind(inf);
q.ReadHMMer3(inf,path);
}
// ... or is infile an old HMMER file?
else if (!strncmp(line,"HMMER",5))
{
if (v>=2) cout<<"Query file is in HMMER format\n";
cerr<<"WARNING: Use of HMMER format as input will result in severe loss of sensitivity!\n";
// Read 'query HMMER file
rewind(inf);
q.ReadHMMer(inf,path);
}
// ... or is it an hhm file?
else if (!strncmp(line,"NAME",4) || !strncmp(line,"HH",2))
{
if (v>=2) cout<<"Query file is in HHM format\n";
// Rewind to beginning of line and read query hhm file
rewind(inf);
q.Read(inf,path);
if (v>=2 && q.Neff_HMM>11.0)
fprintf(stderr,"WARNING: HMM %s looks too diverse (Neff=%.1f>11). Better check the underlying alignment... \n",q.name,q.Neff_HMM);
}
// ... or is it an alignment file
else
{
Alignment* pali;
if (qali==NULL) pali=new(Alignment); else pali=qali;
if (par.calibrate) {
printf("\nError in %s: only HHM files can be calibrated.\n",program_name);
printf("Build an HHM file from your alignment with 'hhmake -i %s' and rerun hhsearch with the hhm file\n\n",infile);
exit(1);
}
if (v>=2 && strcmp(infile,"stdin")) cout<<infile<<" is in A2M, A3M or FASTA format\n";
// Read alignment from infile into matrix X[k][l] as ASCII (and supply first line as extra argument)
pali->Read(inf,infile,line);
// Convert ASCII to int (0-20),throw out all insert states, record their number in I[k][i]
// and store marked sequences in name[k] and seq[k]
pali->Compress(infile);
// Sort out the nseqdis most dissimilar sequences for display in the output alignments
pali->FilterForDisplay(par.max_seqid,par.coverage,par.qid,par.qsc,par.nseqdis);
// Remove sequences with seq. identity larger than seqid percent (remove the shorter of two)
pali->N_filtered = pali->Filter(par.max_seqid,par.coverage,par.qid,par.qsc,par.Ndiff);
if (par.Neff>=0.999)
pali->FilterNeff();
// Calculate pos-specific weights, AA frequencies and transitions -> f[i][a], tr[i][a]
pali->FrequenciesAndTransitions(q);
if (v>=2 && q.Neff_HMM>11.0)
fprintf(stderr,"WARNING: alignment %s looks too diverse (Neff=%.1f>11). Better check it with an alignment viewer... \n",q.name,q.Neff_HMM);
if (qali==NULL) delete(pali);
}
fclose(inf);
return;
}
