1 // This file is part of the Mussa source distribution.
2 // http://mussa.caltech.edu/
3 // Contact author: Tristan De Buysscher, tristan@caltech.edu
5 // This program and all associated source code files are Copyright (C) 2005
6 // the California Institute of Technology, Pasadena, CA, 91125 USA. It is
7 // under the GNU Public License; please see the included LICENSE.txt
8 // file for more information, or contact Tristan directly.
11 // ----------------------------------------
12 // ---------- mussa_class.cc -----------
13 // ----------------------------------------
15 #include <boost/filesystem/operations.hpp>
16 #include <boost/filesystem/fstream.hpp>
17 namespace fs = boost::filesystem;
22 #include "mussa_exceptions.hpp"
23 #include "alg/mussa.hpp"
24 #include "alg/flp.hpp"
32 connect(&the_paths, SIGNAL(progress(const std::string&, int, int)),
33 this, SIGNAL(progress(const std::string&, int, int)));
36 Mussa::Mussa(const Mussa& m)
37 : analysis_name(m.analysis_name),
39 threshold(m.threshold),
40 soft_thres(m.soft_thres),
42 win_append(m.win_append),
43 thres_append(m.thres_append),
44 motif_sequences(m.motif_sequences),
45 color_mapper(m.color_mapper)
47 connect(&the_paths, SIGNAL(progress(const std::string&, int, int)),
48 this, SIGNAL(progress(const std::string&, int, int)));
51 // set all parameters to null state
56 ana_mode = TransitiveNway;
62 motif_sequences.clear();
67 // these 5 simple methods manually set the parameters for doing an analysis
68 // used so that the gui can take input from user and setup the analysis
69 // note - still need a set_append(bool, bool) method...
71 Mussa::set_name(string a_name)
73 analysis_name = a_name;
76 string Mussa::get_name()
84 if (the_seqs.size() > 0)
85 return the_seqs.size();
91 Mussa::set_window(int a_window)
96 int Mussa::get_window() const
102 Mussa::set_threshold(int a_threshold)
104 threshold = a_threshold;
105 if (a_threshold > soft_thres)
106 soft_thres = a_threshold;
109 int Mussa::get_threshold() const
115 Mussa::set_soft_threshold(int new_threshold)
117 if (new_threshold < threshold) {
118 soft_thres = threshold;
119 } else if (new_threshold > window) {
122 soft_thres = new_threshold;
126 int Mussa::get_soft_threshold() const
132 Mussa::set_analysis_mode(enum analysis_modes new_ana_mode)
134 ana_mode = new_ana_mode;
137 enum Mussa::analysis_modes Mussa::get_analysis_mode() const
142 string Mussa::get_analysis_mode_name() const
147 return string("Transitive");
150 return string("Radial");
153 return string("Entropy");
156 return string("[deprecated] Recursive");
159 throw runtime_error("invalid analysis mode type");
164 const NwayPaths& Mussa::paths() const
169 //template <class IteratorT>
170 //void Mussa::createLocalAlignment(IteratorT begin, IteratorT end)
171 void Mussa::createLocalAlignment(std::list<ConservedPath>::iterator begin,
172 std::list<ConservedPath>::iterator end,
173 std::list<ConservedPath::path_type>& result,
174 std::list<std::vector<bool> >& reversed)
176 const vector<Sequence>& raw_seq = the_seqs;
177 ConservedPath::path_type aligned_path;
180 int window_length, win_i;
183 vector<bool> rc_list;
185 vector<bool> matched;
189 for(std::list<ConservedPath>::iterator pathz_i=begin; pathz_i != end; ++pathz_i)
191 ConservedPath& a_path = *pathz_i;
192 window_length = a_path.window_size;
193 // determine which parts of the path are RC relative to first species
194 rc_list = a_path.reverseComplimented();
196 // loop over each bp in the conserved region for all sequences
197 for(win_i = 0; win_i < window_length; win_i++)
199 aligned_path.clear();
200 // determine which exact base pairs match between the sequences
202 for(i2 = 0; i2 < a_path.size()-1; i2++)
204 // assume not rc as most likely, adjust below
207 // no matter the case, any RC node needs adjustments
209 rc_1 = window_length-1;
210 if (a_path[i2+1] < 0)
211 rc_2 = window_length-1;
213 x_start = (abs(a_path[i2]-rc_1+win_i));
214 x_end = (abs(a_path[i2+1]-rc_2+win_i));
217 // ugh, and xor...only want rc coloring if just one of the nodes is rc
218 // if both nodes are rc, then they are 'normal' relative to each other
219 if((rc_list[i2] || rc_list[i2+1] )&&!(rc_list[i2] && rc_list[i2+1]))
220 { //the hideous rc matching logic - not complex, but annoying
221 if(!(( (raw_seq[i2][x_start]=='A')&&(raw_seq[i2+1][x_end]=='T')) ||
222 ((raw_seq[i2][x_start]=='T')&&(raw_seq[i2+1][x_end]=='A')) ||
223 ((raw_seq[i2][x_start]=='G')&&(raw_seq[i2+1][x_end]=='C')) ||
224 ((raw_seq[i2][x_start]=='C')&&(raw_seq[i2+1][x_end]=='G'))) )
228 aligned_path.push_back(x_start);
233 if (!( (raw_seq[i2][x_start] == raw_seq[i2+1][x_end]) &&
234 (raw_seq[i2][x_start] != 'N') &&
235 (raw_seq[i2+1][x_end] != 'N') ) ) {
238 aligned_path.push_back(x_start);
242 // grab the last part of our path, assuming we matched
244 aligned_path.push_back(x_end);
246 if (aligned_path.size() == a_path.size()) {
247 result.push_back(aligned_path);
248 reversed.push_back(rc_list);
256 // takes a string and sets it as the next seq
258 Mussa::add_a_seq(string a_seq)
263 the_seqs.push_back(aSeq);
266 const vector<Sequence>&
267 Mussa::sequences() const
272 void Mussa::load_sequence(fs::path seq_file, fs::path annot_file,
273 int fasta_index, int sub_seq_start, int sub_seq_end)
276 aseq.load_fasta(seq_file, fasta_index, sub_seq_start, sub_seq_end);
277 if ( not annot_file.empty() ) {
278 aseq.load_annot(annot_file, sub_seq_start, sub_seq_end);
280 the_seqs.push_back(aseq);
284 Mussa::load_mupa_file(fs::path para_file_path)
286 fs::ifstream para_file;
287 string file_data_line;
290 int split_index, fasta_index;
291 int sub_seq_start, sub_seq_end;
292 bool seq_params, did_seq;
295 string::size_type new_index, dir_index;
300 // if file was opened, read the parameter values
301 if (fs::exists(para_file_path))
303 para_file.open(para_file_path, ios::in);
305 // what directory is the mupa file in?
306 fs::path file_path_base = para_file_path.branch_path();
308 // setup loop by getting file's first line
309 getline(para_file,file_data_line);
310 split_index = file_data_line.find(" ");
311 param = file_data_line.substr(0,split_index);
312 value = file_data_line.substr(split_index+1);
314 while (!para_file.eof())
317 if (param == "ANA_NAME")
318 analysis_name = value;
319 else if (param == "APPEND_WIN")
321 else if (param == "APPEND_THRES")
323 else if (param == "SEQUENCE_NUM")
324 ; // ignore sequence_num now
325 else if (param == "WINDOW")
326 window = atoi(value.c_str());
327 else if (param == "THRESHOLD")
328 threshold = atoi(value.c_str());
329 else if (param == "SEQUENCE")
331 fs::path seq_file = file_path_base / value;
332 //cout << "seq_file_name " << seq_files.back() << endl;
339 while ((!para_file.eof()) && seq_params)
341 getline(para_file,file_data_line);
342 split_index = file_data_line.find(" ");
343 param = file_data_line.substr(0,split_index);
344 value = file_data_line.substr(split_index+1);
346 if (param == "FASTA_INDEX")
347 fasta_index = atoi(value.c_str());
348 else if (param == "ANNOTATION")
349 annot_file = file_path_base / value;
350 else if (param == "SEQ_START")
351 sub_seq_start = atoi(value.c_str());
352 else if (param == "SEQ_END")
354 sub_seq_end = atoi(value.c_str());
356 //ignore empty lines or that start with '#'
357 else if ((param == "") || (param == "#")) {}
358 else seq_params = false;
360 load_sequence(seq_file, annot_file, fasta_index, sub_seq_start,
364 //ignore empty lines or that start with '#'
365 else if ((param == "") || (param == "#")) {}
368 clog << "Illegal/misplaced mussa parameter in file\n";
369 clog << param << "\n";
374 getline(para_file,file_data_line);
375 split_index = file_data_line.find(" ");
376 param = file_data_line.substr(0,split_index);
377 value = file_data_line.substr(split_index+1);
384 soft_thres = threshold;
385 //cout << "nway mupa: analysis_name = " << analysis_name
386 // << " window = " << window
387 // << " threshold = " << threshold << endl;
389 // no file was loaded, signal error
392 throw mussa_load_error("Config File: " + para_file_path.string() + " not found");
400 time_t t1, t2, begin, end;
401 double seqloadtime, seqcomptime, nwaytime, savetime, totaltime;
407 if (the_seqs.size() < 2) {
408 throw mussa_analysis_error("you need to have at least 2 sequences to "
411 //cout << "nway ana: seq_num = " << the_seqs.size() << endl;
414 seqloadtime = difftime(t2, t1);
419 seqcomptime = difftime(t2, t1);
422 the_paths.setup(window, threshold);
425 nwaytime = difftime(t2, t1);
430 savetime = difftime(t2, t1);
433 totaltime = difftime(end, begin);
435 //cout << "seqload\tseqcomp\tnway\tsave\ttotal\n";
436 //cout << seqloadtime << "\t";
437 //cout << seqcomptime << "\t";
438 //cout << nwaytime << "\t";
439 //cout << savetime << "\t";
440 //cout << totaltime << "\n";
446 vector<int> seq_lens;
447 vector<FLPs> empty_FLP_vector;
449 string save_file_string;
451 empty_FLP_vector.clear();
452 for(vector<Sequence>::size_type i = 0; i < the_seqs.size(); i++)
454 all_comps.push_back(empty_FLP_vector);
455 for(vector<Sequence>::size_type i2 = 0; i2 < the_seqs.size(); i2++)
456 all_comps[i].push_back(dummy_comp);
458 for(vector<Sequence>::size_type i = 0; i < the_seqs.size(); i++) {
459 seq_lens.push_back(the_seqs[i].size());
461 int seqcomps_done = 0;
462 int seqcomps_todo = (the_seqs.size() * (the_seqs.size()-1)) / 2;
463 emit progress("seqcomp", seqcomps_done, seqcomps_todo);
465 for(vector<Sequence>::size_type i = 0; i < the_seqs.size(); i++)
466 for(vector<Sequence>::size_type i2 = i+1; i2 < the_seqs.size(); i2++)
468 //cout << "seqcomping: " << i << " v. " << i2 << endl;
469 all_comps[i][i2].setup(window, threshold);
470 all_comps[i][i2].seqcomp(the_seqs[i].get_seq(), the_seqs[i2].get_seq(), false);
471 all_comps[i][i2].seqcomp(the_seqs[i].get_seq(),the_seqs[i2].rev_comp(),true);
473 emit progress("seqcomp", seqcomps_done, seqcomps_todo);
480 vector<string> some_Seqs;
482 the_paths.set_soft_threshold(soft_thres);
484 if (ana_mode == TransitiveNway) {
485 the_paths.trans_path_search(all_comps);
487 else if (ana_mode == RadialNway) {
488 the_paths.radiate_path_search(all_comps);
490 else if (ana_mode == EntropyNway)
492 //unlike other methods, entropy needs to look at the sequence at this stage
494 for(vector<Sequence>::size_type i = 0; i < the_seqs.size(); i++)
496 some_Seqs.push_back(the_seqs[i].get_seq());
499 the_paths.setup_ent(ent_thres, some_Seqs); // ent analysis extra setup
500 the_paths.entropy_path_search(all_comps);
503 // old recursive transitive analysis function
504 else if (ana_mode == RecursiveNway)
505 the_paths.find_paths_r(all_comps);
507 the_paths.simple_refine();
514 fs::path flp_filepath;
515 fs::fstream save_file;
516 ostringstream append_info;
517 int dir_create_status;
519 if (not analysis_name.empty()) {
520 // not sure why, but gotta close file each time since can't pass
522 save_name = analysis_name;
524 // gotta do bit with adding win & thres if to be appended
528 append_info << "_w" << window;
529 save_name += append_info.str();
535 append_info << "_t" << threshold;
536 save_name += append_info.str();
538 fs::path save_path( save_name);
540 if (not fs::exists(save_path)) {
541 fs::create_directory(save_path);
543 // save sequence and annots to a special mussa file
544 save_file.open(save_path / (save_name+".museq"), ios::out);
545 save_file << "<Mussa_Sequence>" << endl;
547 for(vector<Sequence>::size_type i = 0; i < the_seqs.size(); i++)
549 the_seqs[i].save(save_file);
552 save_file << "</Mussa_Sequence>" << endl;
555 // save nway paths to its mussa save file
556 the_paths.save(save_path / (save_name + ".muway"));
558 for(vector<Sequence>::size_type i = 0; i < the_seqs.size(); i++)
559 for(vector<Sequence>::size_type i2 = i+1; i2 < the_seqs.size(); i2++)
562 append_info << "_sp_" << i << "v" << i2;
563 all_comps[i][i2].save(save_path/(save_name+append_info.str()+".flp"));
569 Mussa::save_muway(fs::path save_path)
571 the_paths.save(save_path);
575 Mussa::load(fs::path ana_file)
578 fs::path file_path_base;
579 fs::path a_file_path;
580 fs::path ana_path(ana_file);
584 ostringstream append_info;
585 vector<FLPs> empty_FLP_vector;
588 //cout << "ana_file name " << ana_file.string() << endl;
589 analysis_name = ana_path.leaf();
590 //cout << " ana_name " << analysis_name << endl;
591 file_path_base = ana_path.branch_path() / analysis_name;
592 a_file_path = file_path_base / (analysis_name + ".muway");
593 //cout << " loading museq: " << a_file_path.string() << endl;
594 the_paths.load(a_file_path);
595 // perhaps this could be more elegent, but at least this'll let
596 // us know what our threshold and window sizes were when we load a muway
597 window = the_paths.get_window();
598 threshold = the_paths.get_threshold();
599 soft_thres = threshold;
601 int seq_num = the_paths.sequence_count();
603 a_file_path = file_path_base / (analysis_name + ".museq");
605 // this is a bit of a hack due to C++ not acting like it should with files
606 for (i = 1; i <= seq_num; i++)
609 //cout << "mussa_class: loading museq frag... " << a_file_path.string() << endl;
610 tmp_seq.load_museq(a_file_path, i);
611 the_seqs.push_back(tmp_seq);
614 empty_FLP_vector.clear();
615 for(i = 0; i < seq_num; i++)
617 all_comps.push_back(empty_FLP_vector);
618 for(i2 = 0; i2 < seq_num; i2++)
619 all_comps[i].push_back(dummy_comp);
622 for(i = 0; i < seq_num; i++)
624 for(i2 = i+1; i2 < seq_num; i2++)
627 append_info << analysis_name << "_sp_" << i << "v" << i2 << ".flp";
628 //cout << append_info.str() << endl;
629 a_file_path = file_path_base / append_info.str();
630 //cout << "path " << a_file_path.string() << endl;
631 all_comps[i][i2].load(a_file_path);
632 //cout << "real size = " << all_comps[i][i2].size() << endl;
641 fs::fstream save_file;
643 save_file.open(analysis_name, ios::out);
645 for(vector<Sequence>::size_type i = 0; i < the_seqs.size(); i++)
646 save_file << the_seqs[i].get_seq() << endl;
648 save_file << window << endl;
650 //note more complex eventually since analysis_name may need to have
651 //window size, threshold and other stuff to modify it...
652 the_paths.save_old(analysis_name);
657 Mussa::load_old(char * load_file_path, int s_num)
660 string file_data_line;
661 int i, space_split_i, comma_split_i;
662 vector<int> loaded_path;
663 string node_pair, node;
667 the_paths.setup(0, 0);
668 save_file.open(load_file_path, ios::in);
670 // currently loads old mussa format
673 for(i = 0; i < seq_num; i++)
675 getline(save_file, file_data_line);
676 a_seq.set_seq(file_data_line);
677 the_seqs.push_back(a_seq);
681 getline(save_file, file_data_line);
682 window = atoi(file_data_line.c_str());
685 while (!save_file.eof())
688 getline(save_file, file_data_line);
689 if (file_data_line != "")
690 for(i = 0; i < seq_num; i++)
692 space_split_i = file_data_line.find(" ");
693 node_pair = file_data_line.substr(0,space_split_i);
694 //cout << "np= " << node_pair;
695 comma_split_i = node_pair.find(",");
696 node = node_pair.substr(comma_split_i+1);
697 //cout << "n= " << node << " ";
698 loaded_path.push_back(atoi (node.c_str()));
699 file_data_line = file_data_line.substr(space_split_i+1);
702 // FIXME: do we have any information about what the threshold should be?
703 the_paths.add_path(0, loaded_path);
707 //the_paths.save("tmp.save");
710 void Mussa::add_motifs(const vector<string>& motifs,
711 const vector<Color>& colors)
713 if (motifs.size() != colors.size()) {
714 throw mussa_error("motif and color vectors must be the same size");
717 for(size_t i = 0; i != motifs.size(); ++i)
719 motif_sequences.insert(motifs[i]);
720 color_mapper.appendInstanceColor("motif", motifs[i], colors[i]);
722 update_sequences_motifs();
725 // I mostly split the ifstream out so I can use a stringstream to test it.
726 void Mussa::load_motifs(std::istream &in)
735 in >> seq >> red >> green >> blue;
736 // if we couldn't read this line 'cause we're like at the end of the file
737 // try to exit the loop
741 seq = Sequence::motif_normalize(seq);
742 } catch(motif_normalize_error e) {
743 clog << "unable to parse " << seq << " skipping" << endl;
744 clog << e.what() << endl;
747 if (red < 0.0 or red > 1.0) {
748 clog << "invalid red value " << red << ". must be in range [0..1]"
752 if (green < 0.0 or green > 1.0) {
753 clog << "invalid green value " << green << ". must be in range [0..1]"
757 if (blue < 0.0 or blue > 1.0) {
758 clog << "invalid blue value " << blue << ". must be in range [0..1]"
762 if (motif_sequences.find(seq) == motif_sequences.end()) {
763 // sequence wasn't found
764 motif_sequences.insert(seq);
765 Color c(red, green, blue);
766 color_mapper.appendInstanceColor("motif", seq, c);
768 clog << "sequence " << seq << " was already defined skipping"
773 update_sequences_motifs();
776 void Mussa::load_motifs(fs::path filename)
779 f.open(filename, ifstream::in);
783 void Mussa::update_sequences_motifs()
785 // once we've loaded all the motifs from the file,
786 // lets attach them to the sequences
787 for(vector<Sequence>::iterator seq_i = the_seqs.begin();
788 seq_i != the_seqs.end();
791 // clear out old motifs
792 seq_i->clear_motifs();
793 // for all the motifs in our set, attach them to the current sequence
794 for(set<string>::iterator motif_i = motif_sequences.begin();
795 motif_i != motif_sequences.end();
798 seq_i->add_motif(*motif_i);
803 const set<string>& Mussa::motifs() const
805 return motif_sequences;
808 AnnotationColors& Mussa::colorMapper()