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 namespace spirit = boost::spirit;
32 #include "alg/sequence.hpp"
33 #include "mussa_exceptions.hpp"
47 annot::annot(int start, int end, std::string type, std::string name)
55 motif::motif(int start, std::string motif)
56 : annot(start, start+motif.size(), "motif", motif),
70 Sequence::Sequence(std::string seq)
72 set_filtered_sequence(seq);
75 Sequence &Sequence::operator=(const Sequence& s)
78 sequence = s.sequence;
86 Sequence &Sequence::operator=(const std::string& s)
88 set_filtered_sequence(s);
92 char Sequence::operator[](int index) const
94 return sequence[index];
97 std::ostream& operator<<(std::ostream& out, const Sequence& seq)
99 out << "Sequence(" << seq.get_seq() << ")";
103 //! load a fasta file into a sequence
105 * \param file_path the location of the fasta file in the filesystem
106 * \param seq_num which sequence in the file to load
107 * \param start_index starting position in the fasta sequence, 0 for beginning
108 * \param end_index ending position in the fasta sequence, 0 for end
109 * \return error message, empty string if no error. (gag!)
112 Sequence::load_fasta(fs::path file_path, int seq_num,
113 int start_index, int end_index)
115 fs::fstream data_file;
116 std::string file_data_line;
117 int header_counter = 0;
118 bool read_seq = true;
119 std::string rev_comp;
120 std::string sequence_raw;
121 std::string seq_tmp; // holds sequence during basic filtering
123 data_file.open(file_path, std::ios::in);
126 throw mussa_load_error("fasta sequence number is 1 based (can't be 0)");
130 throw mussa_load_error("Sequence File: " + file_path.string() + " not found");
132 // if file opened okay, read it
135 // search for the header of the fasta sequence we want
136 while ( (!data_file.eof()) && (header_counter < seq_num) )
138 getline(data_file,file_data_line);
139 if (file_data_line.substr(0,1) == ">")
143 if (header_counter > 0) {
144 header = file_data_line.substr(1);
148 while ( !data_file.eof() && read_seq ) {
149 getline(data_file,file_data_line);
150 if (file_data_line.substr(0,1) == ">")
152 else sequence_raw += file_data_line;
155 // Lastly, if subselection of the sequence was specified we keep cut out
156 // and only keep that part
157 // end_index = 0 means no end was specified, so cut to the end
159 end_index = sequence_raw.size();
161 // sequence filtering for upcasing agctn and convert non AGCTN to N
162 set_filtered_sequence(sequence_raw, start_index, end_index-start_index);
164 std::stringstream errormsg;
165 errormsg << file_path.native_file_string()
166 << " did not have any fasta sequences" << std::endl;
167 throw mussa_load_error(errormsg.str());
173 void Sequence::set_filtered_sequence(const std::string &old_seq,
174 std::string::size_type start,
175 std::string::size_type count)
177 char conversionTable[257];
180 count = old_seq.size() - start;
182 sequence.reserve(count);
184 // Make a conversion table
186 // everything we don't specify below will become 'N'
187 for(int table_i=0; table_i < 256; table_i++)
189 conversionTable[table_i] = 'N';
191 // add end of string character for printing out table for testing purposes
192 conversionTable[256] = '\0';
194 // we want these to map to themselves - ie not to change
195 conversionTable[(int)'A'] = 'A';
196 conversionTable[(int)'T'] = 'T';
197 conversionTable[(int)'G'] = 'G';
198 conversionTable[(int)'C'] = 'C';
200 conversionTable[(int)'a'] = 'A';
201 conversionTable[(int)'t'] = 'T';
202 conversionTable[(int)'g'] = 'G';
203 conversionTable[(int)'c'] = 'C';
205 // finally, the actual conversion loop
206 for(std::string::size_type seq_index = 0; seq_index < count; seq_index++)
208 sequence += conversionTable[ (int)old_seq[seq_index+start]];
212 // this doesn't work properly under gcc 3.x ... it can't recognize toupper
213 //transform(sequence.begin(), sequence.end(), sequence.begin(), toupper);
216 Sequence::load_annot(fs::path file_path, int start_index, int end_index)
218 fs::fstream data_stream(file_path, std::ios::in);
221 throw mussa_load_error("Sequence File: " + file_path.string() + " not found");
224 // so i should probably be passing the parse function some iterators
225 // but the annotations files are (currently) small, so i think i can
226 // get away with loading the whole file into memory
229 while(data_stream.good()) {
235 parse_annot(data, start_index, end_index);
238 /* If this works, yikes, this is some brain hurting code.
240 * what's going on is that when pb_annot is instantiated it stores references
241 * to begin, end, name, type, declared in the parse function, then
242 * when operator() is called it grabs values from those references
243 * and uses that to instantiate an annot object and append that to our
246 * This weirdness is because the spirit library requires that actions
247 * conform to a specific prototype operator()(IteratorT, IteratorT)
248 * which doesn't provide any useful opportunity for me to actually
249 * grab the results of our parsing.
251 * so I instantiate this structure in order to have a place to grab
255 struct push_back_annot {
256 std::list<annot>& annot_list;
262 push_back_annot(std::list<annot>& annot_list_,
267 : annot_list(annot_list_),
275 void operator()(std::string::const_iterator,
276 std::string::const_iterator) const
278 std::cout << "adding annot: " << begin << " " << end << " " << name << " " << type << std::endl;
279 annot_list.push_back(annot(begin, end, name, type));
283 struct push_back_seq {
284 std::list<Sequence>& seq_list;
288 push_back_seq(std::list<Sequence>& seq_list_,
291 : seq_list(seq_list_),
297 void operator()(std::string::const_iterator,
298 std::string::const_iterator) const
300 // filter out newlines from our sequence
302 for(std::string::const_iterator seq_i = seq.begin();
306 if (*seq_i != '\n') new_seq += *seq_i;
308 std::cout << "adding seq: " << name << " " << new_seq << std::endl;
312 seq_list.push_back(s);
317 Sequence::parse_annot(std::string data, int start_index, int end_index)
324 std::list<Sequence> query_seqs;
326 bool status = spirit::parse(data.begin(), data.end(),
329 (+(spirit::alpha_p))[spirit::assign_a(species)] >>
330 +(spirit::space_p) >>
332 ( // parse an absolute location name
333 (spirit::uint_p[spirit::assign_a(start)] >>
335 spirit::uint_p[spirit::assign_a(end)] >>
337 (*(spirit::alpha_p|spirit::digit_p))[spirit::assign_a(name)] >>
341 (*(spirit::alpha_p))[spirit::assign_a(type)]
343 // to understand how this group gets set
344 // read the comment above struct push_back_annot
345 )[push_back_annot(annots, start, end, type, name)]
347 (spirit::ch_p('>') >>
348 (*(~spirit::chlit<char>('\n')))[spirit::assign_a(name)] >>
350 (+(spirit::ch_p('A')|
355 spirit::ch_p('\n')))[spirit::assign_a(seq)]
356 )[push_back_seq(query_seqs, name, seq)]
362 spirit::space_p*/).full;
367 Sequence::load_annot(std::istream& data_stream, int start_index, int end_index)
369 std::string file_data_line;
371 std::string::size_type space_split_i;
372 std::string annot_value;
373 std::list<annot>::iterator list_i;
378 getline(data_stream,file_data_line);
379 species = file_data_line;
381 // end_index = 0 means no end was specified, so cut to the end
383 end_index = sequence.length();
385 //std::cout << "START: " << start_index << " END: " << end_index << std::endl;
387 while ( !data_stream.eof() )
389 getline(data_stream,file_data_line);
390 if (file_data_line != "")
392 // need to get 4 values...almost same code 4 times...
393 // get annot start index
394 space_split_i = file_data_line.find(" ");
395 annot_value = file_data_line.substr(0,space_split_i);
396 an_annot.start = atoi (annot_value.c_str());
397 file_data_line = file_data_line.substr(space_split_i+1);
398 // get annot end index
399 space_split_i = file_data_line.find(" ");
400 annot_value = file_data_line.substr(0,space_split_i);
401 an_annot.end = atoi (annot_value.c_str());
402 file_data_line = file_data_line.substr(space_split_i+1);
404 //std::cout << "seq, annots: " << an_annot.start << ", " << an_annot.end
408 space_split_i = file_data_line.find(" ");
409 if (space_split_i == std::string::npos) // no entries for name & type
411 std::cout << "seq, annots - no name or type\n";
417 annot_value = file_data_line.substr(0,space_split_i);
418 an_annot.name = annot_value;
419 file_data_line = file_data_line.substr(space_split_i+1);
421 space_split_i = file_data_line.find(" ");
422 if (space_split_i == std::string::npos) // no entry for type
426 annot_value = file_data_line.substr(0,space_split_i);
427 an_annot.type = annot_value;
432 // add annot to list if it falls within the range of sequence specified
433 if ((start_index <= an_annot.start) && (end_index >= an_annot.end))
435 an_annot.start -= start_index;
436 an_annot.end -= start_index;
437 annots.push_back(an_annot);
439 // else no (or bogus) annotations
445 const std::string& Sequence::get_species() const
450 bool Sequence::empty() const
452 return (size() == 0);
455 const std::list<annot>& Sequence::annotations() const
460 std::string::size_type Sequence::length() const
465 std::string::size_type Sequence::size() const
467 return sequence.size();
470 Sequence::iterator Sequence::begin()
472 return sequence.begin();
475 Sequence::const_iterator Sequence::begin() const
477 return sequence.begin();
480 Sequence::iterator Sequence::end()
482 return sequence.end();
485 Sequence::const_iterator Sequence::end() const
487 return sequence.end();
492 Sequence::get_seq() const
499 Sequence::subseq(int start, int end) const
501 return sequence.substr(start, end);
506 Sequence::c_seq() const
508 return sequence.c_str();
512 Sequence::rev_comp() const
514 std::string rev_comp;
515 char conversionTable[257];
516 int seq_i, table_i, len;
518 len = sequence.length();
519 rev_comp.reserve(len);
520 // make a conversion table
521 // init all parts of conversion table to '~' character
522 // '~' I doubt will ever appear in a sequence file (jeez, I hope)
523 // and may the fleas of 1000 camels infest the genitals of any biologist (and
524 // seven generations of their progeny) who decides to make it mean
525 // something special!!!
526 // PS - double the curse for any smartass non-biologist who tries it as well
527 for(table_i=0; table_i < 256; table_i++)
529 conversionTable[table_i] = '~';
531 // add end of string character for printing out table for testing purposes
532 conversionTable[256] = '\0';
534 // add in the characters for the bases we want to convert
535 conversionTable[(int)'A'] = 'T';
536 conversionTable[(int)'T'] = 'A';
537 conversionTable[(int)'G'] = 'C';
538 conversionTable[(int)'C'] = 'G';
539 conversionTable[(int)'N'] = 'N';
541 // finally, the actual conversion loop
542 for(seq_i = len - 1; seq_i >= 0; seq_i--)
544 table_i = (int) sequence[seq_i];
545 rev_comp += conversionTable[table_i];
551 void Sequence::set_header(std::string &header_)
557 Sequence::get_header() const
562 //FIXME: i don't think this code is callable
564 Sequence::sp_name() const
571 Sequence::set_seq(const std::string& a_seq)
573 set_filtered_sequence(a_seq);
595 Sequence::save(fs::fstream &save_file)
596 //std::string save_file_path)
599 std::list<annot>::iterator annots_i;
601 // not sure why, or if i'm doing something wrong, but can't seem to pass
602 // file pointers down to this method from the mussa control class
603 // so each call to save a sequence appends to the file started by mussa_class
604 //save_file.open(save_file_path.c_str(), std::ios::app);
606 save_file << "<Sequence>" << std::endl;
607 save_file << sequence << std::endl;
608 save_file << "</Sequence>" << std::endl;
610 save_file << "<Annotations>" << std::endl;
611 save_file << species << std::endl;
612 for (annots_i = annots.begin(); annots_i != annots.end(); ++annots_i)
614 save_file << annots_i->start << " " << annots_i->end << " " ;
615 save_file << annots_i->name << " " << annots_i->type << std::endl;
617 save_file << "</Annotations>" << std::endl;
622 Sequence::load_museq(fs::path load_file_path, int seq_num)
624 fs::fstream load_file;
625 std::string file_data_line;
628 std::string::size_type space_split_i;
629 std::string annot_value;
632 load_file.open(load_file_path, std::ios::in);
635 // search for the seq_num-th sequence
636 while ( (!load_file.eof()) && (seq_counter < seq_num) )
638 getline(load_file,file_data_line);
639 if (file_data_line == "<Sequence>")
642 getline(load_file, file_data_line);
643 sequence = file_data_line;
644 getline(load_file, file_data_line);
645 getline(load_file, file_data_line);
646 if (file_data_line == "<Annotations>")
648 getline(load_file, file_data_line);
649 species = file_data_line;
650 while ( (!load_file.eof()) && (file_data_line != "</Annotations>") )
652 getline(load_file,file_data_line);
653 if ((file_data_line != "") && (file_data_line != "</Annotations>"))
655 // need to get 4 values...almost same code 4 times...
656 // get annot start index
657 space_split_i = file_data_line.find(" ");
658 annot_value = file_data_line.substr(0,space_split_i);
659 an_annot.start = atoi (annot_value.c_str());
660 file_data_line = file_data_line.substr(space_split_i+1);
661 // get annot end index
662 space_split_i = file_data_line.find(" ");
663 annot_value = file_data_line.substr(0,space_split_i);
664 an_annot.end = atoi (annot_value.c_str());
666 if (space_split_i == std::string::npos) // no entry for type or name
668 std::cout << "seq, annots - no type or name\n";
672 else // else get annot type
674 file_data_line = file_data_line.substr(space_split_i+1);
675 space_split_i = file_data_line.find(" ");
676 annot_value = file_data_line.substr(0,space_split_i);
677 an_annot.type = annot_value;
678 if (space_split_i == std::string::npos) // no entry for name
680 std::cout << "seq, annots - no name\n";
683 else // get annot name
685 file_data_line = file_data_line.substr(space_split_i+1);
686 space_split_i = file_data_line.find(" ");
687 annot_value = file_data_line.substr(0,space_split_i);
688 an_annot.type = annot_value;
691 annots.push_back(an_annot); // don't forget to actually add the annot
693 //std::cout << "seq, annots: " << an_annot.start << ", " << an_annot.end
694 // << "-->" << an_annot.type << "::" << an_annot.name << std::endl;
702 Sequence::rc_motif(std::string a_motif)
704 std::string rev_comp;
705 char conversionTable[257];
706 int seq_i, table_i, len;
708 len = a_motif.length();
709 rev_comp.reserve(len);
711 for(table_i=0; table_i < 256; table_i++)
713 conversionTable[table_i] = '~';
715 // add end of std::string character for printing out table for testing purposes
716 conversionTable[256] = '\0';
718 // add in the characters for the bases we want to convert (IUPAC)
719 conversionTable[(int)'A'] = 'T';
720 conversionTable[(int)'T'] = 'A';
721 conversionTable[(int)'G'] = 'C';
722 conversionTable[(int)'C'] = 'G';
723 conversionTable[(int)'N'] = 'N';
724 conversionTable[(int)'M'] = 'K';
725 conversionTable[(int)'R'] = 'Y';
726 conversionTable[(int)'W'] = 'W';
727 conversionTable[(int)'S'] = 'S';
728 conversionTable[(int)'Y'] = 'R';
729 conversionTable[(int)'K'] = 'M';
730 conversionTable[(int)'V'] = 'B';
731 conversionTable[(int)'H'] = 'D';
732 conversionTable[(int)'D'] = 'H';
733 conversionTable[(int)'B'] = 'V';
735 // finally, the actual conversion loop
736 for(seq_i = len - 1; seq_i >= 0; seq_i--)
738 //std::cout << "** i = " << seq_i << " bp = " <<
739 table_i = (int) a_motif[seq_i];
740 rev_comp += conversionTable[table_i];
743 //std::cout << "seq: " << a_motif << std::endl;
744 //std::cout << "rc: " << rev_comp << std::endl;
750 Sequence::motif_normalize(std::string a_motif)
752 std::string valid_motif;
755 len = a_motif.length();
756 valid_motif.reserve(len);
758 // this just upcases IUPAC symbols. Eventually should return an error if non IUPAC is present.
759 // current nonIUPAC symbols are omitted, which is not reported atm
760 for(seq_i = 0; seq_i < len; seq_i++)
762 if ((a_motif[seq_i] == 'a') || (a_motif[seq_i] == 'A'))
764 else if ((a_motif[seq_i] == 't') || (a_motif[seq_i] == 'T'))
766 else if ((a_motif[seq_i] == 'g') || (a_motif[seq_i] == 'G'))
768 else if ((a_motif[seq_i] == 'c') || (a_motif[seq_i] == 'C'))
770 else if ((a_motif[seq_i] == 'n') || (a_motif[seq_i] == 'N'))
772 else if ((a_motif[seq_i] == 'm') || (a_motif[seq_i] == 'M'))
774 else if ((a_motif[seq_i] == 'r') || (a_motif[seq_i] == 'R'))
776 else if ((a_motif[seq_i] == 'w') || (a_motif[seq_i] == 'W'))
778 else if ((a_motif[seq_i] == 's') || (a_motif[seq_i] == 'S'))
780 else if ((a_motif[seq_i] == 'y') || (a_motif[seq_i] == 'Y'))
782 else if ((a_motif[seq_i] == 'k') || (a_motif[seq_i] == 'K'))
784 else if ((a_motif[seq_i] == 'v') || (a_motif[seq_i] == 'V'))
786 else if ((a_motif[seq_i] == 'h') || (a_motif[seq_i] == 'H'))
788 else if ((a_motif[seq_i] == 'd') || (a_motif[seq_i] == 'D'))
790 else if ((a_motif[seq_i] == 'b') || (a_motif[seq_i] == 'B'))
793 std::string msg = "Letter ";
794 msg += a_motif[seq_i];
795 msg += " is not a valid IUPAC symbol";
796 throw motif_normalize_error(msg);
799 //std::cout << "valid_motif is: " << valid_motif << std::endl;
803 void Sequence::add_motif(std::string a_motif)
805 std::vector<int> motif_starts = find_motif(a_motif);
807 for(std::vector<int>::iterator motif_start_i = motif_starts.begin();
808 motif_start_i != motif_starts.end();
811 motif_list.push_back(motif(*motif_start_i, a_motif));
815 void Sequence::clear_motifs()
820 const std::list<motif>& Sequence::motifs() const
826 Sequence::find_motif(std::string a_motif)
828 std::vector<int> motif_match_starts;
829 std::string a_motif_rc;
831 motif_match_starts.clear();
833 //std::cout << "motif is: " << a_motif << std::endl;
834 a_motif = motif_normalize(a_motif);
835 //std::cout << "motif is: " << a_motif << std::endl;
839 //std::cout << "Sequence: none blank motif\n";
840 motif_scan(a_motif, &motif_match_starts);
842 a_motif_rc = rc_motif(a_motif);
843 // make sure not to do search again if it is a palindrome
844 if (a_motif_rc != a_motif)
845 motif_scan(a_motif_rc, &motif_match_starts);
847 return motif_match_starts;
851 Sequence::motif_scan(std::string a_motif, std::vector<int> * motif_match_starts)
854 std::string::size_type seq_i;
855 int motif_i, motif_len;
857 // faster to loop thru the sequence as a old c std::string (ie char array)
858 seq_c = (char*)sequence.c_str();
859 //std::cout << "Sequence: motif, seq len = " << sequence.length() << std::endl;
860 motif_len = a_motif.length();
862 //std::cout << "motif_length: " << motif_len << std::endl;
863 //std::cout << "RAAARRRRR\n";
867 //std::cout << "motif: " << a_motif << std::endl;
869 //std::cout << "Sequence: motif, length= " << length << std::endl;
871 while (seq_i < sequence.length())
873 //std::cout << seq_c[seq_i];
874 //std::cout << seq_c[seq_i] << "?" << a_motif[motif_i] << ":" << motif_i << " ";
875 // this is pretty much a straight translation of Nora's python code
876 // to match iupac letter codes
877 if (a_motif[motif_i] =='N')
879 else if (a_motif[motif_i] == seq_c[seq_i])
881 else if ((a_motif[motif_i] =='M') &&
882 ((seq_c[seq_i]=='A') || (seq_c[seq_i]=='C')))
884 else if ((a_motif[motif_i] =='R') &&
885 ((seq_c[seq_i]=='A') || (seq_c[seq_i]=='G')))
887 else if ((a_motif[motif_i] =='W') &&
888 ((seq_c[seq_i]=='A') || (seq_c[seq_i]=='T')))
890 else if ((a_motif[motif_i] =='S') &&
891 ((seq_c[seq_i]=='C') || (seq_c[seq_i]=='G')))
893 else if ((a_motif[motif_i] =='Y') &&
894 ((seq_c[seq_i]=='C') || (seq_c[seq_i]=='T')))
896 else if ((a_motif[motif_i] =='K') &&
897 ((seq_c[seq_i]=='G') || (seq_c[seq_i]=='T')))
899 else if ((a_motif[motif_i] =='V') &&
900 ((seq_c[seq_i]=='A') || (seq_c[seq_i]=='C') ||
901 (seq_c[seq_i]=='G')))
903 else if ((a_motif[seq_i] =='H') &&
904 ((seq_c[seq_i]=='A') || (seq_c[seq_i]=='C') ||
905 (seq_c[seq_i]=='T')))
907 else if ((a_motif[motif_i] =='D') &&
908 ((seq_c[seq_i]=='A') || (seq_c[seq_i]=='G') ||
909 (seq_c[seq_i]=='T')))
911 else if ((a_motif[motif_i] =='B') &&
912 ((seq_c[seq_i]=='C') || (seq_c[seq_i]=='G') ||
913 (seq_c[seq_i]=='T')))
922 // end Nora stuff, now we see if a match is found this pass
923 if (motif_i == motif_len)
927 motif_match_starts->push_back(seq_i - motif_len + 1);
933 //std::cout << std::endl;
936 void Sequence::add_string_annotation(std::string a_seq,
939 std::vector<int> seq_starts = find_motif(a_seq);
941 for(std::vector<int>::iterator seq_start_i = seq_starts.begin();
942 seq_start_i != seq_starts.end();
945 annots.push_back(annot(*seq_start_i,
946 *seq_start_i+a_seq.size(),
952 void Sequence::find_sequences(std::list<Sequence>::iterator start,
953 std::list<Sequence>::iterator end)
955 while (start != end) {
956 add_string_annotation(start->get_seq(), start->get_header());