// ---------- mussa_class.cc -----------
// ----------------------------------------
+#include <boost/filesystem/path.hpp>
#include <boost/filesystem/operations.hpp>
#include <boost/filesystem/fstream.hpp>
namespace fs = boost::filesystem;
-#include <boost/spirit/core.hpp>
-#include <boost/spirit/actor/push_back_actor.hpp>
-#include <boost/spirit/iterator/file_iterator.hpp>
-#include <boost/spirit/utility/chset.hpp>
-namespace spirit = boost::spirit;
-
#include <iostream>
#include <sstream>
#include "mussa_exceptions.hpp"
-#include "alg/mussa.hpp"
-#include "alg/flp.hpp"
+
+#include "flp.hpp"
+#include "io.hpp"
+#include "mussa.hpp"
+#include "motif_parser.hpp"
using namespace std;
: color_mapper(new AnnotationColors)
{
clear();
- connect(&the_paths, SIGNAL(progress(const std::string&, int, int)),
- this, SIGNAL(progress(const std::string&, int, int)));
+ connect(&the_paths, SIGNAL(progress(const QString&, int, int)),
+ this, SIGNAL(progress(const QString&, int, int)));
}
Mussa::Mussa(const Mussa& m)
analysis_path(m.analysis_path),
dirty(m.dirty)
{
- connect(&the_paths, SIGNAL(progress(const std::string&, int, int)),
- this, SIGNAL(progress(const std::string&, int, int)));
+ connect(&the_paths, SIGNAL(progress(const QString&, int, int)),
+ this, SIGNAL(progress(const QString&, int, int)));
+}
+
+MussaRef Mussa::init()
+{
+ boost::shared_ptr<Mussa> m(new Mussa());
+ return m;
}
boost::filesystem::path Mussa::get_analysis_path() const
return analysis_path;
}
+void Mussa::set_analysis_path(boost::filesystem::path pathname)
+{
+ analysis_path = pathname;
+}
+
// set all parameters to null state
void
Mussa::clear()
the_seqs.clear();
the_paths.clear();
analysis_path = fs::path();
- dirty = false;
+ set_dirty(false);
+}
+
+void Mussa::set_dirty(bool new_state)
+{
+ if (dirty != new_state) {
+ dirty = new_state;
+ emit isModified(dirty);
+ }
}
bool Mussa::is_dirty() const
Mussa::set_name(string a_name)
{
analysis_name = a_name;
- dirty = true;
+ set_dirty(true);
}
-string Mussa::get_name()
+string Mussa::get_name() const
{
return analysis_name;
}
+string Mussa::get_title() const
+{
+ fs::path analysis_path = get_analysis_path();
+ if (not analysis_path.empty()) {
+ return analysis_path.native_file_string();
+ } else if (get_name().size() > 0) {
+ return get_name();
+ } else {
+ return std::string("Unnamed");
+ }
+}
+
int
Mussa::size() const
{
Mussa::set_window(int a_window)
{
window = a_window;
- dirty = true;
+ set_dirty(true);
}
int Mussa::get_window() const
Mussa::set_threshold(int a_threshold)
{
threshold = a_threshold;
- dirty = true;
+ set_dirty(true);
if (a_threshold > soft_thres) {
soft_thres = a_threshold;
}
Mussa::set_analysis_mode(enum analysis_modes new_ana_mode)
{
ana_mode = new_ana_mode;
- dirty = true;
+ set_dirty(true);
}
enum Mussa::analysis_modes Mussa::get_analysis_mode() const
{
boost::shared_ptr<Sequence> seq_copy(new Sequence(a_seq));
the_seqs.push_back(seq_copy);
- dirty = true;
+ set_dirty(true);
}
void Mussa::append_sequence(boost::shared_ptr<Sequence> a_seq)
{
the_seqs.push_back(a_seq);
- dirty = true;
+ set_dirty(true);
}
-const vector<boost::shared_ptr<Sequence> >&
+const vector<SequenceRef>&
Mussa::sequences() const
{
return the_seqs;
aseq->set_species(*name);
}
the_seqs.push_back(aseq);
- dirty = true;
+ set_dirty(true);
+}
+
+void Mussa::load_mupa_file(std::string para_file_path) {
+ load_mupa_file(boost::filesystem::path(para_file_path));
}
void
Mussa::load_mupa_file(fs::path para_file_path)
{
- fs::ifstream para_file;
+ if (not fs::exists(para_file_path))
+ {
+ throw mussa_load_error("Config File: " + para_file_path.string() + " not found");
+ } else if (fs::is_directory(para_file_path)) {
+ throw mussa_load_error("Config File: " + para_file_path.string() + " is a directory.");
+ } else if (fs::is_empty(para_file_path)) {
+ throw mussa_load_error("Config File: " + para_file_path.string() + " is empty");
+ } else {
+ // what directory is the mupa file in?
+ fs::path file_path_base( para_file_path.branch_path()) ;
+
+ fs::ifstream para_file;
+ para_file.open(para_file_path, ios::in);
+
+ load_mupa_stream(para_file, file_path_base);
+ para_file.close();
+ }
+}
+
+void
+Mussa::load_mupa_stream(std::istream& para_file, fs::path& file_path_base)
+{
string file_data_line;
string param, value;
fs::path annot_file;
// initialize values
clear();
- // if file was opened, read the parameter values
- if (not fs::exists(para_file_path))
+ // setup loop by getting file's first line
+ getline(para_file, file_data_line);
+ split_index = file_data_line.find(" ");
+ param = file_data_line.substr(0,split_index);
+ value = file_data_line.substr(split_index+1);
+
+ while (para_file)
{
- throw mussa_load_error("Config File: " + para_file_path.string() + " not found");
- } else if (fs::is_directory(para_file_path)) {
- throw mussa_load_error("Config File: " + para_file_path.string() + " is a directory.");
- } else if (fs::is_empty(para_file_path)) {
- throw mussa_load_error("Config File: " + para_file_path.string() + " is empty");
- } else {
- para_file.open(para_file_path, ios::in);
-
- // what directory is the mupa file in?
- fs::path file_path_base = para_file_path.branch_path();
-
- // setup loop by getting file's first line
- getline(para_file,file_data_line);
- split_index = file_data_line.find(" ");
- param = file_data_line.substr(0,split_index);
- value = file_data_line.substr(split_index+1);
-
- while (para_file)
+ did_seq = false;
+ if (param == "ANA_NAME")
+ analysis_name = value;
+ else if (param == "APPEND_WIN")
+ win_append = true;
+ else if (param == "APPEND_THRES")
+ thres_append = true;
+ else if (param == "SEQUENCE_NUM")
+ ; // ignore sequence_num now
+ else if (param == "WINDOW")
+ window = atoi(value.c_str());
+ else if (param == "THRESHOLD")
+ threshold = atoi(value.c_str());
+ else if (param == "SEQUENCE")
{
- did_seq = false;
- if (param == "ANA_NAME")
- analysis_name = value;
- else if (param == "APPEND_WIN")
- win_append = true;
- else if (param == "APPEND_THRES")
- thres_append = true;
- else if (param == "SEQUENCE_NUM")
- ; // ignore sequence_num now
- else if (param == "WINDOW")
- window = atoi(value.c_str());
- else if (param == "THRESHOLD")
- threshold = atoi(value.c_str());
- else if (param == "SEQUENCE")
- {
- fs::path seq_file = file_path_base / value;
- //cout << "seq_file_name " << seq_files.back() << endl;
- fasta_index = 1;
- annot_file = "";
- sub_seq_start = 0;
- sub_seq_end = 0;
- seq_params = true;
-
- while (para_file && seq_params)
- {
- getline(para_file,file_data_line);
- split_index = file_data_line.find(" ");
- param = file_data_line.substr(0,split_index);
- value = file_data_line.substr(split_index+1);
-
- if (param == "FASTA_INDEX")
- fasta_index = atoi(value.c_str());
- else if (param == "ANNOTATION")
- annot_file = file_path_base / value;
- else if (param == "SEQ_START")
- sub_seq_start = atoi(value.c_str());
- else if (param == "SEQ_END")
- {
- sub_seq_end = atoi(value.c_str());
- }
- //ignore empty lines or that start with '#'
- else if ((param == "") || (param == "#")) {}
- else seq_params = false;
- }
- load_sequence(seq_file, annot_file, fasta_index, sub_seq_start,
- sub_seq_end);
- did_seq = true;
- }
- //ignore empty lines or that start with '#'
- else if ((param == "") || (param == "#")) {}
- else
- {
- clog << "Illegal/misplaced mussa parameter in file\n";
- clog << param << "\n";
- }
-
- if (!did_seq)
+ fs::path seq_file = file_path_base / value;
+ //cout << "seq_file_name " << seq_files.back() << endl;
+ fasta_index = 1;
+ annot_file = "";
+ sub_seq_start = 0;
+ sub_seq_end = 0;
+ seq_params = true;
+
+ while (para_file && seq_params)
{
- getline(para_file,file_data_line);
+ multiplatform_getline(para_file,file_data_line);
split_index = file_data_line.find(" ");
param = file_data_line.substr(0,split_index);
value = file_data_line.substr(split_index+1);
- did_seq = false;
+
+ if (param == "FASTA_INDEX")
+ fasta_index = atoi(value.c_str());
+ else if (param == "ANNOTATION")
+ annot_file = file_path_base / value;
+ else if (param == "SEQ_START")
+ sub_seq_start = atoi(value.c_str());
+ else if (param == "SEQ_END")
+ {
+ sub_seq_end = atoi(value.c_str());
+ }
+ //ignore empty lines or that start with '#'
+ else if ((param == "") || (param == "#")) {
+ // pass
+ } else {
+ seq_params = false;
+ }
}
+ load_sequence(seq_file, annot_file, fasta_index, sub_seq_start,
+ sub_seq_end);
+ did_seq = true;
+ }
+ //ignore empty lines or that start with '#'
+ else if ((param == "") || (param == "#")) {}
+ else
+ {
+ clog << "Illegal/misplaced mussa parameter in file\n";
+ clog << param << "\n";
}
- para_file.close();
-
- soft_thres = threshold;
- //cout << "nway mupa: analysis_name = " << analysis_name
- // << " window = " << window
- // << " threshold = " << threshold << endl;
+ if (!did_seq)
+ {
+ multiplatform_getline(para_file,file_data_line);
+ split_index = file_data_line.find(" ");
+ param = file_data_line.substr(0,split_index);
+ value = file_data_line.substr(split_index+1);
+ did_seq = false;
+ }
}
+
+ soft_thres = threshold;
// no file was loaded, signal error
- dirty = true;
+ set_dirty(true);
}
throw mussa_analysis_error("you need to have at least 2 sequences to "
"do an analysis.");
}
- //cout << "nway ana: seq_num = " << the_seqs.size() << endl;
-
+
seqcomp();
the_paths.setup(window, threshold);
nway();
void
Mussa::save(fs::path save_path)
{
- fs::path flp_filepath;
fs::fstream save_file;
ostringstream append_info;
int dir_create_status;
if (save_path.empty()) {
- if (not analysis_name.empty()) {
+ if (not analysis_path.empty()) {
+ save_path = analysis_path;
+ } else if (not analysis_name.empty()) {
std::string save_name = analysis_name;
// gotta do bit with adding win & thres if to be appended
if (win_append) {
if (not fs::exists(save_path)) {
fs::create_directory(save_path);
}
+
+ std::string basename = save_path.leaf();
+ fs::path museq(basename + ".museq", fs::native);
+
// save sequence and annots to a special mussa file
- save_file.open(save_path / (save_path.leaf()+".museq"), ios::out);
+ save_file.open(save_path / museq, ios::out);
save_file << "<Mussa_Sequence>" << endl;
for(vector<Sequence>::size_type i = 0; i < the_seqs.size(); i++)
save_file << "</Mussa_Sequence>" << endl;
save_file.close();
+ // if we have any motifs, save them.
+ if (motif_sequences.size()) {
+ fs::path mtl(basename + ".mtl", fs::native);
+ save_motifs(save_path / mtl);
+ }
+
// save nway paths to its mussa save file
- the_paths.save(save_path / (save_path.leaf()+ ".muway"));
+ fs::path muway(basename + ".muway", fs::native);
+ the_paths.save(save_path / muway);
for(vector<Sequence>::size_type i = 0; i < the_seqs.size(); i++) {
for(vector<Sequence>::size_type i2 = i+1; i2 < the_seqs.size(); i2++)
{
append_info.str("");
append_info << "_sp_" << i << "v" << i2;
- all_comps[i][i2].save(save_path/(save_path.leaf()+append_info.str()+".flp"));
+ fs::path flp(basename+append_info.str()+".flp", fs::native);
+ all_comps[i][i2].save(save_path / flp);
}
}
- dirty = false;
+
+ set_dirty(false);
analysis_path = save_path;
}
vector<FLPs> empty_FLP_vector;
FLPs dummy_comp;
+
+ //--------------------------------------------------------
+ // Load Muway
+ //--------------------------------------------------------
analysis_path = ana_file;
- //clog << "ana_file name " << ana_file.string() << endl;
analysis_name = ana_path.leaf();
- //clog << " ana_name " << analysis_name << endl;
- file_path_base = ana_path.branch_path() / analysis_name;
- a_file_path = file_path_base / (analysis_name + ".muway");
- //clog << " loading museq: " << a_file_path.string() << endl;
+ fs::path muway(analysis_name+".muway", fs::native);
+ a_file_path = analysis_path / muway;
the_paths.load(a_file_path);
// perhaps this could be more elegent, but at least this'll let
// us know what our threshold and window sizes were when we load a muway
threshold = the_paths.get_threshold();
soft_thres = threshold;
- int seq_num = the_paths.sequence_count();
- a_file_path = file_path_base / (analysis_name + ".museq");
+ //--------------------------------------------------------
+ // Load Sequence
+ //--------------------------------------------------------
+ //int seq_num = the_paths.sequence_count();
+
+ fs::path museq(analysis_name + ".museq", fs::native);
+ a_file_path = analysis_path / museq;
// this is a bit of a hack due to C++ not acting like it should with files
- for (i = 1; i <= seq_num; i++)
+ /*for (i = 1; i <= seq_num; i++)
{
boost::shared_ptr<Sequence> tmp_seq(new Sequence);
- //clog << "mussa_class: loading museq frag... " << a_file_path.string() << endl;
tmp_seq->load_museq(a_file_path, i);
the_seqs.push_back(tmp_seq);
+ }*/
+
+ i = 1;
+ //int seq_num = 0;
+ boost::filesystem::fstream load_museq_fs;
+ load_museq_fs.open(a_file_path, std::ios::in);
+ boost::shared_ptr<Sequence> tmp_seq;
+ while (1)
+ {
+ tmp_seq = Sequence::load_museq(load_museq_fs);
+
+ if (tmp_seq)
+ {
+ the_seqs.push_back(tmp_seq);
+ }
+ else
+ {
+ break;
+ }
+
+
+ //safe guard in case of an infinate loop.
+ //FIXME: If mussa can handle a comparison of 10000 sequences
+ // in the future, then this code should be fixed.
+ if (i == 10000)
+ {
+ throw mussa_load_error(" Run away sequence load!");
+ }
+ i++;
+ }
+ load_museq_fs.close();
+
+ //--------------------------------------------------------
+ // Load Motifs
+ //--------------------------------------------------------
+ fs::path mtl(analysis_name + ".mtl", fs::native);
+ fs::path motif_file = analysis_path / mtl;
+ if (fs::exists(motif_file)) {
+ load_motifs(motif_file);
}
+ vector<Sequence>::size_type seq_num = the_seqs.size();
empty_FLP_vector.clear();
for(i = 0; i < seq_num; i++)
{
all_comps[i].push_back(dummy_comp);
}
+
for(i = 0; i < seq_num; i++)
{
for(i2 = i+1; i2 < seq_num; i2++)
append_info.str("");
append_info << analysis_name << "_sp_" << i << "v" << i2 << ".flp";
//clog << append_info.str() << endl;
- a_file_path = file_path_base / append_info.str();
- //clog << "path " << a_file_path.string() << endl;
+ fs::path flp(append_info.str(), fs::native);
+ a_file_path = analysis_path / flp;
all_comps[i][i2].load(a_file_path);
- //clog << "real size = " << all_comps[i][i2].size() << endl;
}
}
}
-
void
Mussa::save_old()
{
{
motif_sequences.insert(motif);
color_mapper->appendInstanceColor("motif", motif.get_sequence(), color);
+ set_dirty(true);
}
void Mussa::set_motifs(const vector<Sequence>& motifs,
update_sequences_motifs();
}
-
-// Helper functor to append created motifs to our Mussa analysis
-struct push_back_motif {
- std::set<Sequence>& motif_set;
- boost::shared_ptr<AnnotationColors> color_mapper;
- std::string& seq_string;
- std::string& name;
- float& red;
- float& green;
- float& blue;
-
- push_back_motif(std::set<Sequence>& motif_set_,
- boost::shared_ptr<AnnotationColors> color_mapper_,
- std::string& seq_,
- std::string& name_,
- float red_, float green_, float blue_)
- : motif_set(motif_set_),
- color_mapper(color_mapper_),
- seq_string(seq_),
- name(name_),
- red(red_),
- green(green_),
- blue(blue_)
- {
- }
-
- void operator()(std::string::const_iterator,
- std::string::const_iterator) const
- {
- //std::cout << "motif: " << seq_string << "/" << name << endl;
-
- Sequence seq(seq_string);
- // shouldn't we have a better field than "fasta header" and speices?
- seq.set_fasta_header(name);
- // we need to clear the name in case the next motif doesn't have one.
- name.clear();
- // be nice if glsequence was a subclass of sequence so we could
- // just attach colors directly to the motif.
- Color c(red, green, blue);
- color_mapper->appendInstanceColor("motif", seq.c_str(), c);
- motif_set.insert(seq);
- };
-};
-
-// I mostly split the ifstream out so I can use a stringstream to test it.
+void Mussa::load_motifs(fs::path filename)
+{
+ fs::ifstream f;
+ f.open(filename, ifstream::in);
+ load_motifs(f);
+}
+
void Mussa::load_motifs(std::istream &in)
{
std::string data;
- const char *alphabet = Sequence::nucleic_iupac_alphabet.c_str();
- string seq;
- string name;
- float red;
- float green;
- float blue;
+ const char *alphabet = Alphabet::dna_cstr;
+ motif_parser::ParsedMotifs parsed_motifs(motif_sequences, color_mapper);
// slurp our data into a string
std::streamsize bytes_read = 1;
bytes_read = in.readsome(buf, bufsiz);
data.append(buf, buf+bytes_read);
}
- // parse our string
- bool status = spirit::parse(data.begin(), data.end(),
- *(
- (
- (
- (+spirit::chset<>(alphabet))[spirit::assign_a(seq)] >>
- +spirit::space_p
- ) >>
- !(
- (spirit::alpha_p >> *spirit::alnum_p)[spirit::assign_a(name)]
- >> +spirit::space_p
- ) >>
- spirit::real_p[spirit::assign_a(red)] >> +spirit::space_p >>
- spirit::real_p[spirit::assign_a(green)] >> +spirit::space_p >>
- spirit::real_p[spirit::assign_a(blue)] >> +spirit::space_p
- )[push_back_motif(motif_sequences, color_mapper, seq, name, red, green, blue)]
- )).full;
+ parsed_motifs.parse(data);
update_sequences_motifs();
}
-void Mussa::load_motifs(fs::path filename)
+void Mussa::save_motifs(fs::path filename)
{
- fs::ifstream f;
- f.open(filename, ifstream::in);
- load_motifs(f);
+ fs::ofstream out_stream;
+ out_stream.open(filename, ofstream::out);
+ save_motifs(out_stream);
+}
+
+void Mussa::save_motifs(std::ostream& out)
+{
+ for(motif_set::iterator motif_i = motif_sequences.begin();
+ motif_i != motif_sequences.end();
+ ++motif_i)
+ {
+ out << motif_i->get_sequence() << " ";
+ if (motif_i->get_name().size() > 0) {
+ out << "\"" << motif_i->get_name() << "\" ";
+ }
+ out << color_mapper->lookup("motif", motif_i->get_sequence());
+ out << std::endl;
+ }
}
void Mussa::update_sequences_motifs()
{
// once we've loaded all the motifs from the file,
// lets attach them to the sequences
- for(vector<boost::shared_ptr<Sequence> >::iterator seq_i = the_seqs.begin();
+ for(vector<SequenceRef >::iterator seq_i = the_seqs.begin();
seq_i != the_seqs.end();
++seq_i)
{
projects / mussa.git / blobdiff