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 // This file is part of the Mussa source distribution.
12 // http://mussa.caltech.edu/
13 // Contact author: Tristan De Buysscher, tristan@caltech.edu
15 // This program and all associated source code files are Copyright (C) 2005
16 // the California Institute of Technology, Pasadena, CA, 91125 USA. It is
17 // under the GNU Public License; please see the included LICENSE.txt
18 // file for more information, or contact Tristan directly.
21 // ----------------------------------------
22 // ---------- sequence.cc -----------
23 // ----------------------------------------
24 #include <boost/filesystem/fstream.hpp>
25 namespace fs = boost::filesystem;
27 #include <boost/spirit/core.hpp>
28 #include <boost/spirit/actor/push_back_actor.hpp>
29 #include <boost/spirit/iterator/file_iterator.hpp>
30 #include <boost/spirit/utility/chset.hpp>
31 namespace spirit = boost::spirit;
33 #include "alg/sequence.hpp"
34 #include "mussa_exceptions.hpp"
50 annot::annot(int begin, int end, std::string type, std::string name)
62 bool operator==(const annot& left, const annot& right)
64 return ((left.begin== right.begin) and
65 (left.end == right.end) and
66 (left.type == right.type) and
67 (left.name == right.name));
70 motif::motif(int begin, std::string motif)
71 : annot(begin, begin+motif.size(), "motif", motif),
81 Sequence::Sequence(alphabet_ref alphabet_)
94 Sequence::Sequence(const char *seq, alphabet_ref alphabet_)
99 strand(UnknownStrand),
103 set_filtered_sequence(seq, alphabet);
106 Sequence::Sequence(const std::string& seq, alphabet_ref alphabet_)
111 strand(UnknownStrand),
115 set_filtered_sequence(seq, alphabet);
118 Sequence::Sequence(const Sequence& o)
121 alphabet(o.alphabet),
122 seq_start(o.seq_start),
123 seq_count(o.seq_count),
128 motif_list(o.motif_list)
132 Sequence &Sequence::operator=(const Sequence& s)
137 alphabet = s.alphabet;
138 seq_start = s.seq_start;
139 seq_count = s.seq_count;
144 motif_list = s.motif_list;
149 static void multiplatform_getline(std::istream& in, std::string& line)
154 while(in.good() and !(c == '\012' or c == '\015') ) {
158 // if we have cr-lf eat it
160 if (c=='\012' or c == '\015') {
165 void Sequence::load_fasta(fs::path file_path, int seq_num, int start_index, int end_index)
167 load_fasta(file_path, alphabet, seq_num, start_index, end_index);
170 //! load a fasta file into a sequence
171 void Sequence::load_fasta(fs::path file_path, alphabet_ref a,
172 int seq_num, int start_index, int end_index)
174 fs::fstream data_file;
175 data_file.open(file_path, std::ios::in);
177 if (!data_file.good())
179 throw mussa_load_error("Sequence File: "+file_path.string()+" not found");
182 load_fasta(data_file, a, seq_num, start_index, end_index);
183 } catch(sequence_empty_error e) {
184 // there doesn't appear to be any sequence
185 // catch and rethrow to include the filename
186 std::stringstream msg;
187 msg << "The selected sequence in "
188 << file_path.native_file_string()
189 << " appears to be empty";
190 throw sequence_empty_error(msg.str());
191 } catch(sequence_empty_file_error e) {
192 std::stringstream errormsg;
193 errormsg << file_path.native_file_string()
194 << " did not have any fasta sequences" << std::endl;
195 throw sequence_empty_file_error(errormsg.str());
200 void Sequence::load_fasta(std::iostream& file,
201 int seq_num, int start_index, int end_index)
203 load_fasta(file, alphabet, seq_num, start_index, end_index);
207 Sequence::load_fasta(std::iostream& data_file, alphabet_ref a,
209 int start_index, int end_index)
211 std::string file_data_line;
212 int header_counter = 0;
213 bool read_seq = true;
214 std::string rev_comp;
215 std::string sequence_raw;
216 std::string seq_tmp; // holds sequence during basic filtering
217 const Alphabet &alpha = get_alphabet(a);
220 throw mussa_load_error("fasta sequence number is 1 based (can't be 0)");
223 // search for the header of the fasta sequence we want
224 while ( (!data_file.eof()) && (header_counter < seq_num) )
226 multiplatform_getline(data_file, file_data_line);
227 if (file_data_line.substr(0,1) == ">")
231 if (header_counter > 0) {
232 header = file_data_line.substr(1);
236 while ( !data_file.eof() && read_seq ) {
237 multiplatform_getline(data_file,file_data_line);
238 if (file_data_line.substr(0,1) == ">")
241 for (std::string::const_iterator line_i = file_data_line.begin();
242 line_i != file_data_line.end();
245 if(alpha.exists(*line_i)) {
246 sequence_raw += *line_i;
248 throw sequence_invalid_load_error("Unrecognized characters in fasta sequence");
254 // Lastly, if subselection of the sequence was specified we keep cut out
255 // and only keep that part
256 // end_index = 0 means no end was specified, so cut to the end
258 end_index = sequence_raw.size();
260 // sequence filtering for upcasing agctn and convert non AGCTN to N
261 if (end_index-start_index <= 0) {
262 std::string msg("The selected sequence appears to be empty");
263 throw sequence_empty_error(msg);
265 set_filtered_sequence(sequence_raw, a, start_index, end_index-start_index);
267 std::string errormsg("There were no fasta sequences");
268 throw sequence_empty_file_error(errormsg);
272 void Sequence::set_filtered_sequence(const std::string &in_seq,
273 alphabet_ref alphabet_,
278 alphabet = alphabet_;
280 count = in_seq.size() - start;
281 boost::shared_ptr<seq_string> new_seq(new seq_string);
282 new_seq->reserve(count);
284 // finally, the actual conversion loop
285 const Alphabet& alpha_impl = get_alphabet(); // go get one of our actual alphabets
286 std::string::const_iterator seq_i = in_seq.begin()+start;
287 for(size_type i = 0; i != count; ++i, ++seq_i)
289 if (alpha_impl.exists(*seq_i)) {
290 new_seq->append(1, *seq_i);
292 new_seq->append(1, 'N');
303 Sequence::load_annot(fs::path file_path, int start_index, int end_index)
305 fs::fstream data_stream(file_path, std::ios::in);
308 throw mussa_load_error("Sequence File: " + file_path.string() + " not found");
311 // so i should probably be passing the parse function some iterators
312 // but the annotations files are (currently) small, so i think i can
313 // get away with loading the whole file into memory
316 while(data_stream.good()) {
322 parse_annot(data, start_index, end_index);
325 /* If this works, yikes, this is some brain hurting code.
327 * what's going on is that when pb_annot is instantiated it stores references
328 * to begin, end, name, type, declared in the parse function, then
329 * when operator() is called it grabs values from those references
330 * and uses that to instantiate an annot object and append that to our
333 * This weirdness is because the spirit library requires that actions
334 * conform to a specific prototype operator()(IteratorT, IteratorT)
335 * which doesn't provide any useful opportunity for me to actually
336 * grab the results of our parsing.
338 * so I instantiate this structure in order to have a place to grab
342 struct push_back_annot {
343 std::list<annot>& annot_list;
349 push_back_annot(std::list<annot>& annot_list_,
354 : annot_list(annot_list_),
362 void operator()(std::string::const_iterator,
363 std::string::const_iterator) const
365 //std::cout << "adding annot: " << begin << "|" << end << "|" << name << "|" << type << std::endl;
366 annot_list.push_back(annot(begin, end, name, type));
370 struct push_back_seq {
371 std::list<Sequence>& seq_list;
375 push_back_seq(std::list<Sequence>& seq_list_,
378 : seq_list(seq_list_),
384 void operator()(std::string::const_iterator,
385 std::string::const_iterator) const
387 // filter out newlines from our sequence
389 for(std::string::const_iterator seq_i = seq.begin();
393 if (*seq_i != '\015' && *seq_i != '\012') new_seq += *seq_i;
395 //std::cout << "adding seq: " << name << " " << new_seq << std::endl;
398 s.set_fasta_header(name);
399 seq_list.push_back(s);
404 Sequence::parse_annot(std::string data, int start_index, int end_index)
411 std::list<Sequence> query_seqs;
413 bool status = spirit::parse(data.begin(), data.end(),
420 )[spirit::assign_a(species)] >>
424 ( // ignore html tags
425 *(spirit::space_p) >>
427 +(~spirit::ch_p('>')) >>
432 ( // parse an absolute location name
433 (spirit::uint_p[spirit::assign_a(start)] >>
435 spirit::uint_p[spirit::assign_a(end)] >>
440 )[spirit::assign_a(name)] >>
447 )[spirit::assign_a(type)]
449 // to understand how this group gets set
450 // read the comment above struct push_back_annot
451 )[push_back_annot(annots, start, end, type, name)]
453 ((spirit::ch_p('>')|spirit::str_p(">")) >>
454 (*(spirit::print_p))[spirit::assign_a(name)] >>
456 (+(spirit::chset<>(Alphabet::nucleic_alphabet.c_str())))[spirit::assign_a(seq)]
457 )[push_back_seq(query_seqs, name, seq)]
464 // go seearch for query sequences
465 find_sequences(query_seqs.begin(), query_seqs.end());
469 void Sequence::add_annotation(const annot& a)
474 const std::list<annot>& Sequence::annotations() const
480 Sequence::subseq(int start, int count)
487 // there might be an off by one error with start+count > size()
488 if ( count == npos || start+count > size()) {
489 count = size()-start;
491 Sequence new_seq(*this);
492 new_seq.parent = this;
493 new_seq.seq_start = seq_start+start;
494 new_seq.seq_count = count;
496 new_seq.motif_list = motif_list;
497 new_seq.annots.clear();
498 // attempt to copy & reannotate position based annotations
499 int end = start+count;
501 for(std::list<annot>::const_iterator annot_i = annots.begin();
502 annot_i != annots.end();
505 int annot_begin= annot_i->begin;
506 int annot_end = annot_i->end;
508 if (annot_begin < end) {
509 if (annot_begin >= start) {
510 annot_begin -= start;
515 if (annot_end < end) {
521 annot new_annot(annot_begin, annot_end, annot_i->type, annot_i->name);
522 new_seq.annots.push_back(new_annot);
529 std::string Sequence::create_reverse_map() const
531 std::string rc_map(256, '~');
532 // if we're rna, use U instead of T
533 // we might want to add an "is_rna" to sequence at somepoint
534 char TU = (alphabet == reduced_rna_alphabet) ? 'U' : 'T';
535 char tu = (alphabet == reduced_rna_alphabet) ? 'u' : 't';
536 rc_map['A'] = TU ; rc_map['a'] = tu ;
537 rc_map['T'] = 'A'; rc_map['t'] = 'a';
538 rc_map['U'] = 'A'; rc_map['u'] = 'a';
539 rc_map['G'] = 'C'; rc_map['g'] = 'c';
540 rc_map['C'] = 'G'; rc_map['c'] = 'g';
541 rc_map['M'] = 'K'; rc_map['m'] = 'k';
542 rc_map['R'] = 'Y'; rc_map['r'] = 'y';
543 rc_map['W'] = 'W'; rc_map['w'] = 'w';
544 rc_map['S'] = 'S'; rc_map['s'] = 's';
545 rc_map['Y'] = 'R'; rc_map['y'] = 'r';
546 rc_map['K'] = 'M'; rc_map['k'] = 'm';
547 rc_map['V'] = 'B'; rc_map['v'] = 'b';
548 rc_map['H'] = 'D'; rc_map['h'] = 'd';
549 rc_map['D'] = 'H'; rc_map['d'] = 'h';
550 rc_map['B'] = 'V'; rc_map['b'] = 'v';
551 rc_map['N'] = 'N'; rc_map['n'] = 'n';
552 rc_map['X'] = 'X'; rc_map['x'] = 'x';
556 rc_map['~'] = '~'; // not really needed, but perhaps it's clearer.
560 Sequence Sequence::rev_comp() const
562 std::string rev_comp;
563 rev_comp.reserve(length());
565 std::string rc_map = create_reverse_map();
567 // reverse and convert
568 seq_string::const_reverse_iterator seq_i;
569 seq_string::const_reverse_iterator seq_end = seq->rend();
570 for(seq_i = seq->rbegin();
574 rev_comp.append(1, rc_map[*seq_i]);
576 return Sequence(rev_comp, alphabet);
579 void Sequence::set_fasta_header(std::string header_)
584 void Sequence::set_species(const std::string& name)
589 std::string Sequence::get_species() const
596 Sequence::get_fasta_header() const
602 Sequence::get_name() const
604 if (header.size() > 0)
606 else if (species.size() > 0)
612 const Alphabet& Sequence::get_alphabet() const
614 return get_alphabet(alphabet);
617 const Alphabet& Sequence::get_alphabet(alphabet_ref alpha) const
620 case reduced_dna_alphabet:
621 return Alphabet::reduced_dna_alphabet;
622 case reduced_rna_alphabet:
623 return Alphabet::reduced_rna_alphabet;
624 case reduced_nucleic_alphabet:
625 return Alphabet::reduced_nucleic_alphabet;
626 case nucleic_alphabet:
627 return Alphabet::nucleic_alphabet;
628 case protein_alphabet:
629 return Alphabet::protein_alphabet;
631 throw std::runtime_error("unrecognized alphabet type");
636 void Sequence::set_sequence(const std::string& s, alphabet_ref a)
638 set_filtered_sequence(s, a);
641 std::string Sequence::get_sequence() const
646 return std::string();
649 Sequence::const_reference Sequence::operator[](Sequence::size_type i) const
654 Sequence::const_reference Sequence::at(Sequence::size_type i) const
656 if (!seq) throw std::out_of_range("empty sequence");
657 return seq->at(i+seq_start);
667 strand = UnknownStrand;
674 const char *Sequence::c_str() const
677 return seq->c_str()+seq_start;
682 Sequence::const_iterator Sequence::begin() const
684 if (seq and seq_count != 0)
685 return seq->begin()+seq_start;
687 return Sequence::const_iterator(0);
690 Sequence::const_iterator Sequence::end() const
692 if (seq and seq_count != 0) {
693 return seq->begin() + seq_start + seq_count;
695 return Sequence::const_iterator(0);
699 bool Sequence::empty() const
701 return (seq_count == 0) ? true : false;
704 Sequence::size_type Sequence::start() const
707 return seq_start - parent->start();
712 Sequence::size_type Sequence::stop() const
714 return start() + seq_count;
717 Sequence::size_type Sequence::size() const
722 Sequence::size_type Sequence::length() const
728 Sequence::save(fs::fstream &save_file)
731 std::list<annot>::iterator annots_i;
733 // not sure why, or if i'm doing something wrong, but can't seem to pass
734 // file pointers down to this method from the mussa control class
735 // so each call to save a sequence appends to the file started by mussa_class
736 //save_file.open(save_file_path.c_str(), std::ios::app);
738 save_file << "<Sequence>" << std::endl;
739 save_file << *this << std::endl;
740 save_file << "</Sequence>" << std::endl;
742 save_file << "<Annotations>" << std::endl;
743 save_file << species << std::endl;
744 for (annots_i = annots.begin(); annots_i != annots.end(); ++annots_i)
746 save_file << annots_i->begin << " " << annots_i->end << " " ;
747 save_file << annots_i->name << " " << annots_i->type << std::endl;
749 save_file << "</Annotations>" << std::endl;
754 Sequence::load_museq(fs::path load_file_path, int seq_num)
756 fs::fstream load_file;
757 std::string file_data_line;
760 std::string::size_type space_split_i;
761 std::string annot_value;
764 load_file.open(load_file_path, std::ios::in);
767 // search for the seq_num-th sequence
768 while ( (!load_file.eof()) && (seq_counter < seq_num) )
770 getline(load_file,file_data_line);
771 if (file_data_line == "<Sequence>")
774 getline(load_file, file_data_line);
775 // looks like the sequence is written as a single line
776 set_filtered_sequence(file_data_line, reduced_dna_alphabet);
777 getline(load_file, file_data_line);
778 getline(load_file, file_data_line);
779 if (file_data_line == "<Annotations>")
781 getline(load_file, file_data_line);
782 species = file_data_line;
783 while ( (!load_file.eof()) && (file_data_line != "</Annotations>") )
785 getline(load_file,file_data_line);
786 if ((file_data_line != "") && (file_data_line != "</Annotations>"))
788 // need to get 4 values...almost same code 4 times...
789 // get annot start index
790 space_split_i = file_data_line.find(" ");
791 annot_value = file_data_line.substr(0,space_split_i);
792 an_annot.begin = atoi (annot_value.c_str());
793 file_data_line = file_data_line.substr(space_split_i+1);
794 // get annot end index
795 space_split_i = file_data_line.find(" ");
796 annot_value = file_data_line.substr(0,space_split_i);
797 an_annot.end = atoi (annot_value.c_str());
799 if (space_split_i == std::string::npos) // no entry for type or name
801 std::cout << "seq, annots - no type or name\n";
805 else // else get annot type
807 file_data_line = file_data_line.substr(space_split_i+1);
808 space_split_i = file_data_line.find(" ");
809 annot_value = file_data_line.substr(0,space_split_i);
810 an_annot.type = annot_value;
811 if (space_split_i == std::string::npos) // no entry for name
813 std::cout << "seq, annots - no name\n";
816 else // get annot name
818 file_data_line = file_data_line.substr(space_split_i+1);
819 space_split_i = file_data_line.find(" ");
820 annot_value = file_data_line.substr(0,space_split_i);
821 an_annot.type = annot_value;
824 annots.push_back(an_annot); // don't forget to actually add the annot
826 //std::cout << "seq, annots: " << an_annot.start << ", " << an_annot.end
827 // << "-->" << an_annot.type << "::" << an_annot.name << std::endl;
834 void Sequence::add_motif(const Sequence& a_motif)
836 std::vector<int> motif_starts = find_motif(a_motif);
838 for(std::vector<int>::iterator motif_start_i = motif_starts.begin();
839 motif_start_i != motif_starts.end();
842 motif_list.push_back(motif(*motif_start_i, a_motif.get_sequence()));
846 void Sequence::clear_motifs()
851 const std::list<motif>& Sequence::motifs() const
857 Sequence::find_motif(const Sequence& a_motif) const
859 std::vector<int> motif_match_starts;
860 Sequence norm_motif_rc;
862 motif_match_starts.clear();
863 // std::cout << "motif is: " << norm_motif << std::endl;
865 if (a_motif.size() > 0)
867 //std::cout << "Sequence: none blank motif\n";
868 motif_scan(a_motif, &motif_match_starts);
870 norm_motif_rc = a_motif.rev_comp();;
871 // make sure not to do search again if it is a palindrome
872 if (norm_motif_rc != a_motif) {
873 motif_scan(norm_motif_rc, &motif_match_starts);
876 return motif_match_starts;
880 Sequence::motif_scan(const Sequence& a_motif, std::vector<int> * motif_match_starts) const
882 // if there's no sequence we can't scan for it?
883 // should this throw an exception?
886 std::string::size_type seq_i = 0;
887 Sequence::size_type motif_i = 0;
888 Sequence::size_type motif_len = a_motif.length();
889 Sequence::value_type motif_char;
890 Sequence::value_type seq_char;
892 while (seq_i < seq->length())
894 // this is pretty much a straight translation of Nora's python code
895 // to match iupac letter codes
896 motif_char = toupper(a_motif[motif_i]);
897 seq_char = toupper(seq->at(seq_i));
898 if (motif_char =='N')
900 else if (motif_char == seq_char)
902 else if ((motif_char =='M') && ((seq_char=='A') || (seq_char=='C')))
904 else if ((motif_char =='R') && ((seq_char=='A') || (seq_char=='G')))
906 else if ((motif_char =='W') && ((seq_char=='A') || (seq_char=='T')))
908 else if ((motif_char =='S') && ((seq_char=='C') || (seq_char=='G')))
910 else if ((motif_char =='Y') && ((seq_char=='C') || (seq_char=='T')))
912 else if ((motif_char =='K') && ((seq_char=='G') || (seq_char=='T')))
914 else if ((motif_char =='V') &&
915 ((seq_char=='A') || (seq_char=='C') || (seq_char=='G')))
917 else if ((motif_char =='H') &&
918 ((seq_char=='A') || (seq_char=='C') || (seq_char=='T')))
920 else if ((motif_char =='D') &&
921 ((seq_char=='A') || (seq_char=='G') || (seq_char=='T')))
923 else if ((motif_char =='B') &&
924 ((seq_char=='C') || (seq_char=='G') || (seq_char=='T')))
928 // if a motif doesn't match, erase our current trial and try again
933 // end Nora stuff, now we see if a match is found this pass
934 if (motif_i == motif_len)
937 motif_match_starts->push_back(seq_i - motif_len + 1);
943 //std::cout << std::endl;
946 void Sequence::add_string_annotation(std::string a_seq,
949 std::vector<int> seq_starts = find_motif(a_seq);
951 //std::cout << "searching for " << a_seq << " found " << seq_starts.size() << std::endl;
953 for(std::vector<int>::iterator seq_start_i = seq_starts.begin();
954 seq_start_i != seq_starts.end();
957 annots.push_back(annot(*seq_start_i,
958 *seq_start_i+a_seq.size(),
964 void Sequence::find_sequences(std::list<Sequence>::iterator start,
965 std::list<Sequence>::iterator end)
967 while (start != end) {
968 add_string_annotation(start->get_sequence(), start->get_fasta_header());
974 std::ostream& operator<<(std::ostream& out, const Sequence& s)
976 for(Sequence::const_iterator s_i = s.begin(); s_i != s.end(); ++s_i) {
982 bool operator<(const Sequence& x, const Sequence& y)
984 Sequence::const_iterator x_i = x.begin();
985 Sequence::const_iterator y_i = y.begin();
986 // for sequences there's some computation associated with computing .end
988 Sequence::const_iterator xend = x.end();
989 Sequence::const_iterator yend = y.end();
991 if( x_i == xend and y_i == yend ) {
993 } else if ( x_i == xend ) {
995 } else if ( y_i == yend ) {
997 } else if ( (*x_i) < (*y_i)) {
999 } else if ( (*x_i) > (*y_i) ) {
1008 bool operator==(const Sequence& x, const Sequence& y)
1010 if (x.empty() and y.empty()) {
1011 // if there's no sequence in either sequence structure, they're equal
1013 } else if (x.empty() or y.empty()) {
1014 // if we fail the first test, and we discover one is empty,
1015 // we know they can't be equal. (and we need to do this
1016 // to prevent dereferencing an empty pointer)
1018 } else if (x.seq_count != y.seq_count) {
1019 // if they're of different lenghts, they're not equal
1022 Sequence::const_iterator xseq_i = x.begin();
1023 Sequence::const_iterator yseq_i = y.begin();
1024 // since the length of the two sequences is equal, we only need to
1026 for(; xseq_i != x.end(); ++xseq_i, ++yseq_i) {
1027 if (toupper(*xseq_i) != toupper(*yseq_i)) {
1034 bool operator!=(const Sequence& x, const Sequence& y)
1036 return not operator==(x, y);