void HMM::load(istream &is)
{
order=0;
// Load number of states and allocate arrays of that size
is>>numStates;
discardInput(is,"states");
emissionProb.resize(numStates);
transitionProb.resize(numStates,numStates);
// Load schema
discardInput(is,"schema");
loadSchema(is);
const int numDiscrete=schema.getNumDiscrete();
discreteEmitProb.resize(numStates,numDiscrete);
chains.resize(numDiscrete);
orders.resize(numStates,numDiscrete);
// Read transition && emission probabilities
discardInput(is,"transitions");
is >> transitionProb;
String line;
for(int q=1 ; q<numStates ; ++q) {
discardInput(is,"emissions");
emissionProb[q].load(is);
for(int i=0 ; i<numDiscrete ; ++i) {
int order;
is>>order;
orders[q][i]=order;
if(order>this->order) this->order=order;
discardInput(is,"order");
discardInput(is,"alphabet");
Alphabet alphabet;
alphabet.load(is);
schema.getAlphabet(i)=alphabet;
HigherOrderAlphabet H(alphabet,order+1);
const int numNmers=H.getNumNmers();
Array1D<double> &row=discreteEmitProb[q][i];
row.resize(numNmers);
for(NmerSymbol nmer=1 ; nmer<numNmers ; ++nmer) {
line.getline(is);
line.trimWhitespace();
if(is.eof() || line.isEmpty()) break;
BOOM::Vector<BOOM::String> &fields=*line.getFields();
Sequence seq(fields[0],alphabet);
NmerSymbol nmer=H.lookup(seq);
row[nmer]=fields[1].asDouble(); // in log space
delete &fields;
}
}
}
// Load list of foreground states (if present)
while(!is.eof()) {
line.getline(is);
line.trimWhitespace();
if(line.contains("foreground")) {
line.getline(is);
line.trimWhitespace();
foregroundStates.clear();
BOOM::Vector<BOOM::String> &fg=*line.getFields(",");
int n=fg.size();
for(int i=0 ; i<n ; ++i) foregroundStates+=fg[i].asInt();
delete &fg;
}
}
// Construct chains
for(int i=0 ; i<numDiscrete ; ++i) {
HigherOrderAlphabet H(schema.getAlphabet(i),order+1);
chains[i]=NmerChain(H);
}
// Convert to log space
logifyTransitions();
}
int main(int argc, char** argv)
{
// prints more information if the program was terminated due to an
// exception.
// see http://groups.yahoo.com/group/open-source-programmer/message/91
#ifdef __GNUC__
std::set_terminate (__gnu_cxx::__verbose_terminate_handler);
#endif
// parse command line arguments
handleargs(argc, argv);
//load an alphabet from the mass_file
Alphabet alphabet;
try {
alphabet.load(alphabet_filename);
}
catch (IOException) {
cerr << "Error loading alphabet \""<< alphabet_filename << "\"" << endl;
exit(1);
}
cout << "Alphabet succesfully loaded:" << endl;
cout << alphabet << endl;
// copy over masses into a alphabet with fast access by character
CharacterAlphabet fast_alphabet(alphabet);
//now build some kind of fragmentizer
cout << "Start building Fragmenter..." << endl;
auto_ptr<Modifier<peaklist_type> > sort_modifier(new SortModifier<peaklist_type>);
auto_ptr<Modifier<peaklist_type> > unification_modifier(new UnificationModifier<peaklist_type>);
auto_ptr<MultiModifier<peaklist_type> > multi_modifier(new MultiModifier<peaklist_type>);
multi_modifier->addModifier(sort_modifier);
multi_modifier->addModifier(unification_modifier);
fragmenter_type fragmenter(alphabet, cleavage_chars, prohibition_chars, with_cleavage_char);
fragmenter.setModifier(auto_ptr<Modifier<peaklist_type> >(multi_modifier));
cout << "...done\n" << endl;
//now that we hopefully have some fragmentizer, we should be able to start parsing the fasta file
//first open the stream
ifstream fasta_stream(database_filename.c_str());
if (!fasta_stream) {
cerr << "Error: file \"" << database_filename << "\" could not be opened" << endl;
exit(1);
}
//some vars
//this maxlength thingie determines, how long such a sequence should be. 0 means unlimited
const unsigned int maxlength = 0;
string line;
string currentString("");
string currentID("");
//open file for output
ofstream ofs(output_filename.c_str());
if (!ofs) {
cerr << "Error: Could not open output file \""<< output_filename << "\"" << endl;
exit(1);
}
//init the statistics vars
long discarded_sequences = 0;
long kept_sequences = 0;
map_type letter_frequencies;
//begin the looping
while (getline(fasta_stream, line)) {
if (line[0] == ' ') {
continue; //FIXME this is not fasta file format
}
if(line[0] == '>') { // check if end of sequence (rather beginng of new)
// close current sequence (if there is one)
if((maxlength==0 || currentString.length()<=maxlength) && currentString.length()>0) {
//here comes the sequence handling
//FIXME we need string toUpper, but we dont have it
if (handle_sequence (currentString, currentID, fast_alphabet, fragmenter, letter_frequencies, ofs))
++kept_sequences;
else
++discarded_sequences;
currentString ="";
}
// begin new id sequence
int k=1;
while(isspace(line[k++]));
currentID="";
for(unsigned int i=k-1;i<line.length() && !(isspace(line[i]));i++) {
currentID += line[i];
}
}
else {
// read sequence line
for(unsigned int i=0;i<line.length();i++) {
if(!isspace(line[i])) {
currentString += line[i];
}
}
}
}
// handle last sequence
if((maxlength==0 || currentString.length()<=maxlength) && currentString.length()>0) {
//here we have to do some handling for the last sequence
//FIXME we need string toUpper, but we dont have it
if (handle_sequence (currentString, currentID, fast_alphabet, fragmenter, letter_frequencies, ofs))
++kept_sequences;
else
++discarded_sequences;
}
//logging
cout << "\nAll sequences processed\n" << endl;
//close streams
fasta_stream.close();
ofs.close();
if (statistic_filename!="") {
//do some statistics for the letter frequencies and sequence counts
ofstream statistics(statistic_filename.c_str());
if (!statistics) {
cerr << "Error opening output file for statistics \""<< statistic_filename << "\"" << endl;
exit(1);
}
cout << "Writing statistics to \"" << statistic_filename << "\"" << endl;
statistics << "# Kept " << kept_sequences << ", discarded " << discarded_sequences << ", from a total of "
<< kept_sequences + discarded_sequences << " sequences." << endl;
statistics << "# Letter frequencies in valid sequences:" << endl;
map_type::const_iterator it = letter_frequencies.begin();
for ( ; it != letter_frequencies.end(); ++it) {
statistics << it->first << "\t" << it->second << endl;
}
statistics.close();
}
}
int main(int argc, char **argv) {
Options options;
try {
options.parse(argc, argv);
} catch (TCLAP::ArgException &e) {
throw ims::Exception("Error while parsing command line: " + e.error() + " for argument " + e.argId());
}
Alphabet alphabet;
try {
alphabet.load(options.getAlphabetFileName());
} catch (IOException& ioe) {
cerr << "can not read alphabet: " << ioe.what() << endl;
return 1;
}
Weights weights(alphabet.getMasses(), 1 /*precision*/);
// optimize alphabet by dividing by gcd
weights.divideByGCD();
// get masses from -m options and/or from a file
vector<double> masses = options.getMasses();
if (options.hasMassFile()) {
MassesTextParser parser;
try {
parser.load(options.getMassFile());
} catch (IOException& ioe) {
cerr << "can not read masses file: " << ioe.what() << endl;
return 1;
}
vector<double> filemasses = parser.getElements();
// append to masses
copy(filemasses.begin(), filemasses.end(), back_inserter(masses));
}
if (masses.empty()) {
cerr << "No input masses given to decompose (use -m or -f option)!\n";
return 1;
}
// maximal number of decompositions to show
unsigned int maxNumber = options.getMaxNumberDecompositions();
printHeader(alphabet, weights, options.getMode());
if (options.getMode() == Options::GETNUMBER) {
ClassicalDPMassDecomposer<> decomposer(weights);
// loop through masses
for (vector<double>::const_iterator it = masses.begin(); it != masses.end(); ++it) {
value_type mass = static_cast<value_type>(*it);
decomposition_value_t number = decomposer.getNumberOfDecompositions(mass);
cout << "# mass " << mass << " has " << number << " decompositions" << endl;
}
} else { // if any other tasks defined
// use decomposer with residues
IntegerMassDecomposer<> decomposer(weights);
if (options.getMode() == Options::FINDALL) {
// loop through masses
for (vector<double>::const_iterator it = masses.begin(); it != masses.end(); ++it) {
value_type mass = static_cast<value_type>(*it);
outputDecompositions(decomposer.getAllDecompositions(mass), alphabet, mass, maxNumber);
cout << '\n';
}
}
else if (options.getMode() == Options::FINDONE) {
// loop through masses
for (vector<double>::const_iterator it = masses.begin(); it != masses.end(); ++it) {
value_type mass = static_cast<value_type>(*it);
decomposition_t decomposition = decomposer.getDecomposition(mass);
if (decomposition.size() == 0) {
cout << "# mass " << mass << " has 0 decompositions\n\n";
} else {
cout << "# mass " << mass << " has at least this decomposition:\n";
printDecomposition(alphabet, decomposition);
cout << "\n\n";
}
}
}
else if (options.getMode() == Options::ISDECOMPOSABLE) {
// loop through masses
for (vector<double>::const_iterator it = masses.begin(); it != masses.end(); ++it) {
value_type mass = static_cast<value_type>(*it);
if (decomposer.exist(mass)) {
cout << "# mass " << mass << " has at least one decomposition\n";
} else {
cout << "# mass " << mass << " has no decompositions\n";
}
}
}
}
cout << "# done\n";
return 0;
}
