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 void Mussa::append_sequence(Sequence a_seq)
258 the_seqs.push_back(a_seq);
261 const vector<Sequence>&
262 Mussa::sequences() const
267 void Mussa::load_sequence(fs::path seq_file, fs::path annot_file,
268 int fasta_index, int sub_seq_start, int sub_seq_end)
271 aseq.load_fasta(seq_file, fasta_index, sub_seq_start, sub_seq_end);
272 if ( not annot_file.empty() ) {
273 aseq.load_annot(annot_file, sub_seq_start, sub_seq_end);
275 the_seqs.push_back(aseq);
279 Mussa::load_mupa_file(fs::path para_file_path)
281 fs::ifstream para_file;
282 string file_data_line;
285 int split_index, fasta_index;
286 int sub_seq_start, sub_seq_end;
287 bool seq_params, did_seq;
290 string::size_type new_index, dir_index;
295 // if file was opened, read the parameter values
296 if (fs::exists(para_file_path))
298 para_file.open(para_file_path, ios::in);
300 // what directory is the mupa file in?
301 fs::path file_path_base = para_file_path.branch_path();
303 // setup loop by getting file's first line
304 getline(para_file,file_data_line);
305 split_index = file_data_line.find(" ");
306 param = file_data_line.substr(0,split_index);
307 value = file_data_line.substr(split_index+1);
309 while (!para_file.eof())
312 if (param == "ANA_NAME")
313 analysis_name = value;
314 else if (param == "APPEND_WIN")
316 else if (param == "APPEND_THRES")
318 else if (param == "SEQUENCE_NUM")
319 ; // ignore sequence_num now
320 else if (param == "WINDOW")
321 window = atoi(value.c_str());
322 else if (param == "THRESHOLD")
323 threshold = atoi(value.c_str());
324 else if (param == "SEQUENCE")
326 fs::path seq_file = file_path_base / value;
327 //cout << "seq_file_name " << seq_files.back() << endl;
334 while ((!para_file.eof()) && seq_params)
336 getline(para_file,file_data_line);
337 split_index = file_data_line.find(" ");
338 param = file_data_line.substr(0,split_index);
339 value = file_data_line.substr(split_index+1);
341 if (param == "FASTA_INDEX")
342 fasta_index = atoi(value.c_str());
343 else if (param == "ANNOTATION")
344 annot_file = file_path_base / value;
345 else if (param == "SEQ_START")
346 sub_seq_start = atoi(value.c_str());
347 else if (param == "SEQ_END")
349 sub_seq_end = atoi(value.c_str());
351 //ignore empty lines or that start with '#'
352 else if ((param == "") || (param == "#")) {}
353 else seq_params = false;
355 load_sequence(seq_file, annot_file, fasta_index, sub_seq_start,
359 //ignore empty lines or that start with '#'
360 else if ((param == "") || (param == "#")) {}
363 clog << "Illegal/misplaced mussa parameter in file\n";
364 clog << param << "\n";
369 getline(para_file,file_data_line);
370 split_index = file_data_line.find(" ");
371 param = file_data_line.substr(0,split_index);
372 value = file_data_line.substr(split_index+1);
379 soft_thres = threshold;
380 //cout << "nway mupa: analysis_name = " << analysis_name
381 // << " window = " << window
382 // << " threshold = " << threshold << endl;
384 // no file was loaded, signal error
387 throw mussa_load_error("Config File: " + para_file_path.string() + " not found");
395 time_t t1, t2, begin, end;
396 double seqloadtime, seqcomptime, nwaytime, savetime, totaltime;
402 if (the_seqs.size() < 2) {
403 throw mussa_analysis_error("you need to have at least 2 sequences to "
406 //cout << "nway ana: seq_num = " << the_seqs.size() << endl;
409 seqloadtime = difftime(t2, t1);
414 seqcomptime = difftime(t2, t1);
417 the_paths.setup(window, threshold);
420 nwaytime = difftime(t2, t1);
425 savetime = difftime(t2, t1);
428 totaltime = difftime(end, begin);
430 //cout << "seqload\tseqcomp\tnway\tsave\ttotal\n";
431 //cout << seqloadtime << "\t";
432 //cout << seqcomptime << "\t";
433 //cout << nwaytime << "\t";
434 //cout << savetime << "\t";
435 //cout << totaltime << "\n";
441 vector<int> seq_lens;
442 vector<FLPs> empty_FLP_vector;
444 string save_file_string;
446 empty_FLP_vector.clear();
447 for(vector<Sequence>::size_type i = 0; i < the_seqs.size(); i++)
449 all_comps.push_back(empty_FLP_vector);
450 for(vector<Sequence>::size_type i2 = 0; i2 < the_seqs.size(); i2++)
451 all_comps[i].push_back(dummy_comp);
453 for(vector<Sequence>::size_type i = 0; i < the_seqs.size(); i++) {
454 seq_lens.push_back(the_seqs[i].size());
456 int seqcomps_done = 0;
457 int seqcomps_todo = (the_seqs.size() * (the_seqs.size()-1)) / 2;
458 emit progress("seqcomp", seqcomps_done, seqcomps_todo);
460 for(vector<Sequence>::size_type i = 0; i < the_seqs.size(); i++)
461 for(vector<Sequence>::size_type i2 = i+1; i2 < the_seqs.size(); i2++)
463 //cout << "seqcomping: " << i << " v. " << i2 << endl;
464 all_comps[i][i2].setup(window, threshold);
465 all_comps[i][i2].seqcomp(the_seqs[i], the_seqs[i2], false);
466 all_comps[i][i2].seqcomp(the_seqs[i], the_seqs[i2].rev_comp(),true);
468 emit progress("seqcomp", seqcomps_done, seqcomps_todo);
475 vector<string> some_Seqs;
477 the_paths.set_soft_threshold(soft_thres);
479 if (ana_mode == TransitiveNway) {
480 the_paths.trans_path_search(all_comps);
482 else if (ana_mode == RadialNway) {
483 the_paths.radiate_path_search(all_comps);
485 else if (ana_mode == EntropyNway)
487 //unlike other methods, entropy needs to look at the sequence at this stage
489 for(vector<Sequence>::size_type i = 0; i < the_seqs.size(); i++)
491 some_Seqs.push_back(the_seqs[i]);
494 the_paths.setup_ent(ent_thres, some_Seqs); // ent analysis extra setup
495 the_paths.entropy_path_search(all_comps);
498 // old recursive transitive analysis function
499 else if (ana_mode == RecursiveNway)
500 the_paths.find_paths_r(all_comps);
502 the_paths.simple_refine();
509 fs::path flp_filepath;
510 fs::fstream save_file;
511 ostringstream append_info;
512 int dir_create_status;
514 if (not analysis_name.empty()) {
515 // not sure why, but gotta close file each time since can't pass
517 save_name = analysis_name;
519 // gotta do bit with adding win & thres if to be appended
523 append_info << "_w" << window;
524 save_name += append_info.str();
530 append_info << "_t" << threshold;
531 save_name += append_info.str();
533 fs::path save_path( save_name);
535 if (not fs::exists(save_path)) {
536 fs::create_directory(save_path);
538 // save sequence and annots to a special mussa file
539 save_file.open(save_path / (save_name+".museq"), ios::out);
540 save_file << "<Mussa_Sequence>" << endl;
542 for(vector<Sequence>::size_type i = 0; i < the_seqs.size(); i++)
544 the_seqs[i].save(save_file);
547 save_file << "</Mussa_Sequence>" << endl;
550 // save nway paths to its mussa save file
551 the_paths.save(save_path / (save_name + ".muway"));
553 for(vector<Sequence>::size_type i = 0; i < the_seqs.size(); i++)
554 for(vector<Sequence>::size_type i2 = i+1; i2 < the_seqs.size(); i2++)
557 append_info << "_sp_" << i << "v" << i2;
558 all_comps[i][i2].save(save_path/(save_name+append_info.str()+".flp"));
564 Mussa::save_muway(fs::path save_path)
566 the_paths.save(save_path);
570 Mussa::load(fs::path ana_file)
573 fs::path file_path_base;
574 fs::path a_file_path;
575 fs::path ana_path(ana_file);
579 ostringstream append_info;
580 vector<FLPs> empty_FLP_vector;
583 //cout << "ana_file name " << ana_file.string() << endl;
584 analysis_name = ana_path.leaf();
585 //cout << " ana_name " << analysis_name << endl;
586 file_path_base = ana_path.branch_path() / analysis_name;
587 a_file_path = file_path_base / (analysis_name + ".muway");
588 //cout << " loading museq: " << a_file_path.string() << endl;
589 the_paths.load(a_file_path);
590 // perhaps this could be more elegent, but at least this'll let
591 // us know what our threshold and window sizes were when we load a muway
592 window = the_paths.get_window();
593 threshold = the_paths.get_threshold();
594 soft_thres = threshold;
596 int seq_num = the_paths.sequence_count();
598 a_file_path = file_path_base / (analysis_name + ".museq");
600 // this is a bit of a hack due to C++ not acting like it should with files
601 for (i = 1; i <= seq_num; i++)
604 //cout << "mussa_class: loading museq frag... " << a_file_path.string() << endl;
605 tmp_seq.load_museq(a_file_path, i);
606 the_seqs.push_back(tmp_seq);
609 empty_FLP_vector.clear();
610 for(i = 0; i < seq_num; i++)
612 all_comps.push_back(empty_FLP_vector);
613 for(i2 = 0; i2 < seq_num; i2++)
614 all_comps[i].push_back(dummy_comp);
617 for(i = 0; i < seq_num; i++)
619 for(i2 = i+1; i2 < seq_num; i2++)
622 append_info << analysis_name << "_sp_" << i << "v" << i2 << ".flp";
623 //cout << append_info.str() << endl;
624 a_file_path = file_path_base / append_info.str();
625 //cout << "path " << a_file_path.string() << endl;
626 all_comps[i][i2].load(a_file_path);
627 //cout << "real size = " << all_comps[i][i2].size() << endl;
636 fs::fstream save_file;
638 save_file.open(analysis_name, ios::out);
640 for(vector<Sequence>::size_type i = 0; i < the_seqs.size(); i++)
641 save_file << the_seqs[i] << endl;
643 save_file << window << endl;
645 //note more complex eventually since analysis_name may need to have
646 //window size, threshold and other stuff to modify it...
647 the_paths.save_old(analysis_name);
652 Mussa::load_old(char * load_file_path, int s_num)
655 string file_data_line;
656 int i, space_split_i, comma_split_i;
657 vector<int> loaded_path;
658 string node_pair, node;
662 the_paths.setup(0, 0);
663 save_file.open(load_file_path, ios::in);
665 // currently loads old mussa format
668 for(i = 0; i < seq_num; i++)
670 getline(save_file, file_data_line);
671 a_seq = file_data_line;
672 the_seqs.push_back(a_seq);
676 getline(save_file, file_data_line);
677 window = atoi(file_data_line.c_str());
680 while (!save_file.eof())
683 getline(save_file, file_data_line);
684 if (file_data_line != "")
685 for(i = 0; i < seq_num; i++)
687 space_split_i = file_data_line.find(" ");
688 node_pair = file_data_line.substr(0,space_split_i);
689 //cout << "np= " << node_pair;
690 comma_split_i = node_pair.find(",");
691 node = node_pair.substr(comma_split_i+1);
692 //cout << "n= " << node << " ";
693 loaded_path.push_back(atoi (node.c_str()));
694 file_data_line = file_data_line.substr(space_split_i+1);
697 // FIXME: do we have any information about what the threshold should be?
698 the_paths.add_path(0, loaded_path);
702 //the_paths.save("tmp.save");
705 void Mussa::add_motifs(const vector<string>& motifs,
706 const vector<Color>& colors)
708 if (motifs.size() != colors.size()) {
709 throw mussa_error("motif and color vectors must be the same size");
712 for(size_t i = 0; i != motifs.size(); ++i)
714 motif_sequences.insert(motifs[i]);
715 color_mapper.appendInstanceColor("motif", motifs[i], colors[i]);
717 update_sequences_motifs();
720 // I mostly split the ifstream out so I can use a stringstream to test it.
721 void Mussa::load_motifs(std::istream &in)
730 in >> seq >> red >> green >> blue;
731 // if we couldn't read this line 'cause we're like at the end of the file
732 // try to exit the loop
736 seq = Sequence::motif_normalize(seq);
737 } catch(motif_normalize_error e) {
738 clog << "unable to parse " << seq << " skipping" << endl;
739 clog << e.what() << endl;
742 if (red < 0.0 or red > 1.0) {
743 clog << "invalid red value " << red << ". must be in range [0..1]"
747 if (green < 0.0 or green > 1.0) {
748 clog << "invalid green value " << green << ". must be in range [0..1]"
752 if (blue < 0.0 or blue > 1.0) {
753 clog << "invalid blue value " << blue << ". must be in range [0..1]"
757 if (motif_sequences.find(seq) == motif_sequences.end()) {
758 // sequence wasn't found
759 motif_sequences.insert(seq);
760 Color c(red, green, blue);
761 color_mapper.appendInstanceColor("motif", seq, c);
763 clog << "sequence " << seq << " was already defined skipping"
768 update_sequences_motifs();
771 void Mussa::load_motifs(fs::path filename)
774 f.open(filename, ifstream::in);
778 void Mussa::update_sequences_motifs()
780 // once we've loaded all the motifs from the file,
781 // lets attach them to the sequences
782 for(vector<Sequence>::iterator seq_i = the_seqs.begin();
783 seq_i != the_seqs.end();
786 // clear out old motifs
787 seq_i->clear_motifs();
788 // for all the motifs in our set, attach them to the current sequence
789 for(set<string>::iterator motif_i = motif_sequences.begin();
790 motif_i != motif_sequences.end();
793 seq_i->add_motif(*motif_i);
798 const set<string>& Mussa::motifs() const
800 return motif_sequences;
803 AnnotationColors& Mussa::colorMapper()