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"
40 // some standard dna alphabets
43 // this should make our sequence parsing end-of-line convention
45 static const char* dna_alphabet = "AaCcGgTtNn\012\015";
46 static const char* rna_alphabet = "AaCcGgNnUu\012\015";
47 static const char* iupac_alphabet = "AaCcGgTtUuRrYyMmKkSsWwBbDdHhVvNn\012\015";
57 annot::annot(int start, int end, std::string type, std::string name)
69 bool operator==(const annot& left, const annot& right)
71 return ((left.start == right.start) and
72 (left.end == right.end) and
73 (left.type == right.type) and
74 (left.name == right.name));
77 motif::motif(int start, std::string motif)
78 : annot(start, start+motif.size(), "motif", motif),
100 Sequence::Sequence(const char *seq)
102 set_filtered_sequence(seq);
105 Sequence::Sequence(const std::string& seq)
107 set_filtered_sequence(seq);
110 Sequence::Sequence(const Sequence& o)
115 motif_list(o.motif_list)
119 Sequence &Sequence::operator=(const Sequence& s)
122 //sequence = s.sequence;
127 motif_list = s.motif_list;
132 static void multiplatform_getline(std::istream& in, std::string& line)
137 while(in.good() and !(c == '\012' or c == '\015') ) {
141 // if we have cr-lf eat it
143 if (c=='\012' or c == '\015') {
148 //! load a fasta file into a sequence
150 * \param file_path the location of the fasta file in the filesystem
151 * \param seq_num which sequence in the file to load
152 * \param start_index starting position in the fasta sequence, 0 for beginning
153 * \param end_index ending position in the fasta sequence, 0 for end
154 * \return error message, empty string if no error. (gag!)
156 void Sequence::load_fasta(fs::path file_path, int seq_num,
157 int start_index, int end_index)
159 fs::fstream data_file;
160 data_file.open(file_path, std::ios::in);
162 if (!data_file.good())
164 throw mussa_load_error("Sequence File: "+file_path.string()+" not found");
167 load_fasta(data_file, seq_num, start_index, end_index);
168 } catch(sequence_empty_error e) {
169 // there doesn't appear to be any sequence
170 // catch and rethrow to include the filename
171 std::stringstream msg;
172 msg << "The selected sequence in "
173 << file_path.native_file_string()
174 << " appears to be empty";
175 throw sequence_empty_error(msg.str());
176 } catch(sequence_empty_file_error e) {
177 std::stringstream errormsg;
178 errormsg << file_path.native_file_string()
179 << " did not have any fasta sequences" << std::endl;
180 throw sequence_empty_file_error(errormsg.str());
186 Sequence::load_fasta(std::iostream& data_file, int seq_num,
187 int start_index, int end_index)
189 std::string file_data_line;
190 int header_counter = 0;
191 bool read_seq = true;
192 std::string rev_comp;
193 std::string sequence_raw;
194 std::string seq_tmp; // holds sequence during basic filtering
197 throw mussa_load_error("fasta sequence number is 1 based (can't be 0)");
200 // search for the header of the fasta sequence we want
201 while ( (!data_file.eof()) && (header_counter < seq_num) )
203 multiplatform_getline(data_file, file_data_line);
204 if (file_data_line.substr(0,1) == ">")
208 if (header_counter > 0) {
209 header = file_data_line.substr(1);
213 while ( !data_file.eof() && read_seq ) {
214 multiplatform_getline(data_file,file_data_line);
215 if (file_data_line.substr(0,1) == ">")
217 else sequence_raw += file_data_line;
220 // Lastly, if subselection of the sequence was specified we keep cut out
221 // and only keep that part
222 // end_index = 0 means no end was specified, so cut to the end
224 end_index = sequence_raw.size();
226 // sequence filtering for upcasing agctn and convert non AGCTN to N
227 if (end_index-start_index <= 0) {
228 std::string msg("The selected sequence appears to be empty");
229 throw sequence_empty_error(msg);
231 set_filtered_sequence(sequence_raw, start_index, end_index-start_index);
233 std::string errormsg("There were no fasta sequences");
234 throw sequence_empty_file_error(errormsg);
238 void Sequence::set_filtered_sequence(const std::string &old_seq,
239 std::string::size_type start,
240 std::string::size_type count)
242 char conversionTable[257];
245 count = old_seq.size() - start;
246 std::string::clear();
249 // Make a conversion table
251 // everything we don't specify below will become 'N'
252 for(int table_i=0; table_i < 256; table_i++)
254 conversionTable[table_i] = 'N';
256 // add end of string character for printing out table for testing purposes
257 conversionTable[256] = '\0';
259 // we want these to map to themselves - ie not to change
260 conversionTable[(int)'A'] = 'A';
261 conversionTable[(int)'T'] = 'T';
262 conversionTable[(int)'G'] = 'G';
263 conversionTable[(int)'C'] = 'C';
265 conversionTable[(int)'a'] = 'A';
266 conversionTable[(int)'t'] = 'T';
267 conversionTable[(int)'g'] = 'G';
268 conversionTable[(int)'c'] = 'C';
270 // finally, the actual conversion loop
271 for(std::string::size_type seq_index = 0; seq_index < count; seq_index++)
273 append(1, conversionTable[ (int)old_seq[seq_index+start]]);
277 // this doesn't work properly under gcc 3.x ... it can't recognize toupper
278 //transform(sequence.begin(), sequence.end(), sequence.begin(), toupper);
281 Sequence::load_annot(fs::path file_path, int start_index, int end_index)
283 fs::fstream data_stream(file_path, std::ios::in);
286 throw mussa_load_error("Sequence File: " + file_path.string() + " not found");
289 // so i should probably be passing the parse function some iterators
290 // but the annotations files are (currently) small, so i think i can
291 // get away with loading the whole file into memory
294 while(data_stream.good()) {
300 parse_annot(data, start_index, end_index);
303 /* If this works, yikes, this is some brain hurting code.
305 * what's going on is that when pb_annot is instantiated it stores references
306 * to begin, end, name, type, declared in the parse function, then
307 * when operator() is called it grabs values from those references
308 * and uses that to instantiate an annot object and append that to our
311 * This weirdness is because the spirit library requires that actions
312 * conform to a specific prototype operator()(IteratorT, IteratorT)
313 * which doesn't provide any useful opportunity for me to actually
314 * grab the results of our parsing.
316 * so I instantiate this structure in order to have a place to grab
320 struct push_back_annot {
321 std::list<annot>& annot_list;
327 push_back_annot(std::list<annot>& annot_list_,
332 : annot_list(annot_list_),
340 void operator()(std::string::const_iterator,
341 std::string::const_iterator) const
343 //std::cout << "adding annot: " << begin << "|" << end << "|" << name << "|" << type << std::endl;
344 annot_list.push_back(annot(begin, end, name, type));
348 struct push_back_seq {
349 std::list<Sequence>& seq_list;
353 push_back_seq(std::list<Sequence>& seq_list_,
356 : seq_list(seq_list_),
362 void operator()(std::string::const_iterator,
363 std::string::const_iterator) const
365 // filter out newlines from our sequence
367 for(std::string::const_iterator seq_i = seq.begin();
371 if (*seq_i != '\015' && *seq_i != '\012') new_seq += *seq_i;
373 //std::cout << "adding seq: " << name << " " << new_seq << std::endl;
377 seq_list.push_back(s);
382 Sequence::parse_annot(std::string data, int start_index, int end_index)
389 std::list<Sequence> query_seqs;
391 bool status = spirit::parse(data.begin(), data.end(),
394 (+(spirit::alpha_p))[spirit::assign_a(species)] >>
395 +(spirit::space_p) >>
397 ( // parse an absolute location name
398 (spirit::uint_p[spirit::assign_a(start)] >>
400 spirit::uint_p[spirit::assign_a(end)] >>
405 )[spirit::assign_a(name)] >>
412 )[spirit::assign_a(type)]
414 // to understand how this group gets set
415 // read the comment above struct push_back_annot
416 )[push_back_annot(annots, start, end, type, name)]
418 (spirit::ch_p('>') >>
419 (*(spirit::print_p))[spirit::assign_a(name)] >>
421 (+(spirit::chset<>(iupac_alphabet)))[spirit::assign_a(seq)]
422 )[push_back_seq(query_seqs, name, seq)]
428 spirit::space_p*/).full;
430 // go seearch for query sequences
431 find_sequences(query_seqs.begin(), query_seqs.end());
434 const std::string& Sequence::get_species() const
439 void Sequence::add_annotation(const annot& a)
444 const std::list<annot>& Sequence::annotations() const
450 Sequence::subseq(int start, int count) const
452 // there might be an off by one error with start+count > size()
453 if ( count == npos || start+count > size()) {
454 count = size()-start;
456 Sequence new_seq(std::string::substr(start, count));
457 new_seq.set_header(get_header());
458 //new_seq.set_species(get_species());
460 new_seq.motif_list = motif_list;
461 // attempt to copy & reannotate position based annotations
462 int end = start+count;
464 for(std::list<annot>::const_iterator annot_i = annots.begin();
465 annot_i != annots.end();
468 int annot_start = annot_i->start;
469 int annot_end = annot_i->end;
471 if (annot_start < end) {
472 if (annot_start >=start) {
473 annot_start -= start;
478 if (annot_end < end) {
484 annot new_annot(annot_start, annot_end, annot_i->type, annot_i->name);
485 new_seq.annots.push_back(new_annot);
493 Sequence::rev_comp() const
495 std::string rev_comp;
496 char conversionTable[257];
497 int seq_i, table_i, len;
500 rev_comp.reserve(len);
501 // make a conversion table
502 // init all parts of conversion table to '~' character
503 // '~' I doubt will ever appear in a sequence file (jeez, I hope)
504 // and may the fleas of 1000 camels infest the genitals of any biologist (and
505 // seven generations of their progeny) who decides to make it mean
506 // something special!!!
507 // PS - double the curse for any smartass non-biologist who tries it as well
508 for(table_i=0; table_i < 256; table_i++)
510 conversionTable[table_i] = '~';
512 // add end of string character for printing out table for testing purposes
513 conversionTable[256] = '\0';
515 // add in the characters for the bases we want to convert
516 conversionTable[(int)'A'] = 'T';
517 conversionTable[(int)'T'] = 'A';
518 conversionTable[(int)'G'] = 'C';
519 conversionTable[(int)'C'] = 'G';
520 conversionTable[(int)'N'] = 'N';
522 // finally, the actual conversion loop
523 for(seq_i = len - 1; seq_i >= 0; seq_i--)
525 table_i = (int) at(seq_i);
526 rev_comp += conversionTable[table_i];
532 void Sequence::set_header(std::string header_)
538 Sequence::get_header() const
543 //FIXME: i don't think this code is callable
545 Sequence::sp_name() const
562 std::string::clear();
569 Sequence::save(fs::fstream &save_file)
570 //std::string save_file_path)
573 std::list<annot>::iterator annots_i;
575 // not sure why, or if i'm doing something wrong, but can't seem to pass
576 // file pointers down to this method from the mussa control class
577 // so each call to save a sequence appends to the file started by mussa_class
578 //save_file.open(save_file_path.c_str(), std::ios::app);
580 save_file << "<Sequence>" << std::endl;
581 save_file << *this << std::endl;
582 save_file << "</Sequence>" << std::endl;
584 save_file << "<Annotations>" << std::endl;
585 save_file << species << std::endl;
586 for (annots_i = annots.begin(); annots_i != annots.end(); ++annots_i)
588 save_file << annots_i->start << " " << annots_i->end << " " ;
589 save_file << annots_i->name << " " << annots_i->type << std::endl;
591 save_file << "</Annotations>" << std::endl;
596 Sequence::load_museq(fs::path load_file_path, int seq_num)
598 fs::fstream load_file;
599 std::string file_data_line;
602 std::string::size_type space_split_i;
603 std::string annot_value;
606 load_file.open(load_file_path, std::ios::in);
609 // search for the seq_num-th sequence
610 while ( (!load_file.eof()) && (seq_counter < seq_num) )
612 getline(load_file,file_data_line);
613 if (file_data_line == "<Sequence>")
616 getline(load_file, file_data_line);
617 assign(file_data_line);
618 getline(load_file, file_data_line);
619 getline(load_file, file_data_line);
620 if (file_data_line == "<Annotations>")
622 getline(load_file, file_data_line);
623 species = file_data_line;
624 while ( (!load_file.eof()) && (file_data_line != "</Annotations>") )
626 getline(load_file,file_data_line);
627 if ((file_data_line != "") && (file_data_line != "</Annotations>"))
629 // need to get 4 values...almost same code 4 times...
630 // get annot start index
631 space_split_i = file_data_line.find(" ");
632 annot_value = file_data_line.substr(0,space_split_i);
633 an_annot.start = atoi (annot_value.c_str());
634 file_data_line = file_data_line.substr(space_split_i+1);
635 // get annot end index
636 space_split_i = file_data_line.find(" ");
637 annot_value = file_data_line.substr(0,space_split_i);
638 an_annot.end = atoi (annot_value.c_str());
640 if (space_split_i == std::string::npos) // no entry for type or name
642 std::cout << "seq, annots - no type or name\n";
646 else // else get annot type
648 file_data_line = file_data_line.substr(space_split_i+1);
649 space_split_i = file_data_line.find(" ");
650 annot_value = file_data_line.substr(0,space_split_i);
651 an_annot.type = annot_value;
652 if (space_split_i == std::string::npos) // no entry for name
654 std::cout << "seq, annots - no name\n";
657 else // get annot name
659 file_data_line = file_data_line.substr(space_split_i+1);
660 space_split_i = file_data_line.find(" ");
661 annot_value = file_data_line.substr(0,space_split_i);
662 an_annot.type = annot_value;
665 annots.push_back(an_annot); // don't forget to actually add the annot
667 //std::cout << "seq, annots: " << an_annot.start << ", " << an_annot.end
668 // << "-->" << an_annot.type << "::" << an_annot.name << std::endl;
676 Sequence::rc_motif(std::string a_motif)
678 std::string rev_comp;
679 char conversionTable[257];
680 int seq_i, table_i, len;
682 len = a_motif.length();
683 rev_comp.reserve(len);
685 for(table_i=0; table_i < 256; table_i++)
687 conversionTable[table_i] = '~';
689 // add end of std::string character for printing out table for testing purposes
690 conversionTable[256] = '\0';
692 // add in the characters for the bases we want to convert (IUPAC)
693 conversionTable[(int)'A'] = 'T';
694 conversionTable[(int)'T'] = 'A';
695 conversionTable[(int)'G'] = 'C';
696 conversionTable[(int)'C'] = 'G';
697 conversionTable[(int)'N'] = 'N';
698 conversionTable[(int)'M'] = 'K';
699 conversionTable[(int)'R'] = 'Y';
700 conversionTable[(int)'W'] = 'W';
701 conversionTable[(int)'S'] = 'S';
702 conversionTable[(int)'Y'] = 'R';
703 conversionTable[(int)'K'] = 'M';
704 conversionTable[(int)'V'] = 'B';
705 conversionTable[(int)'H'] = 'D';
706 conversionTable[(int)'D'] = 'H';
707 conversionTable[(int)'B'] = 'V';
709 // finally, the actual conversion loop
710 for(seq_i = len - 1; seq_i >= 0; seq_i--)
712 //std::cout << "** i = " << seq_i << " bp = " <<
713 table_i = (int) a_motif[seq_i];
714 rev_comp += conversionTable[table_i];
717 //std::cout << "seq: " << a_motif << std::endl;
718 //std::cout << "rc: " << rev_comp << std::endl;
724 Sequence::motif_normalize(std::string a_motif)
726 std::string valid_motif;
729 len = a_motif.length();
730 valid_motif.reserve(len);
732 // this just upcases IUPAC symbols. Eventually should return an error if non IUPAC is present.
733 // current nonIUPAC symbols are omitted, which is not reported atm
734 for(seq_i = 0; seq_i < len; seq_i++)
736 if ((a_motif[seq_i] == 'a') || (a_motif[seq_i] == 'A'))
738 else if ((a_motif[seq_i] == 't') || (a_motif[seq_i] == 'T'))
740 else if ((a_motif[seq_i] == 'g') || (a_motif[seq_i] == 'G'))
742 else if ((a_motif[seq_i] == 'c') || (a_motif[seq_i] == 'C'))
744 else if ((a_motif[seq_i] == 'n') || (a_motif[seq_i] == 'N'))
746 else if ((a_motif[seq_i] == 'm') || (a_motif[seq_i] == 'M'))
748 else if ((a_motif[seq_i] == 'r') || (a_motif[seq_i] == 'R'))
750 else if ((a_motif[seq_i] == 'w') || (a_motif[seq_i] == 'W'))
752 else if ((a_motif[seq_i] == 's') || (a_motif[seq_i] == 'S'))
754 else if ((a_motif[seq_i] == 'y') || (a_motif[seq_i] == 'Y'))
756 else if ((a_motif[seq_i] == 'k') || (a_motif[seq_i] == 'K'))
758 else if ((a_motif[seq_i] == 'v') || (a_motif[seq_i] == 'V'))
760 else if ((a_motif[seq_i] == 'h') || (a_motif[seq_i] == 'H'))
762 else if ((a_motif[seq_i] == 'd') || (a_motif[seq_i] == 'D'))
764 else if ((a_motif[seq_i] == 'b') || (a_motif[seq_i] == 'B'))
767 std::string msg = "Letter ";
768 msg += a_motif[seq_i];
769 msg += " is not a valid IUPAC symbol";
770 throw motif_normalize_error(msg);
773 //std::cout << "valid_motif is: " << valid_motif << std::endl;
777 void Sequence::add_motif(std::string a_motif)
779 std::vector<int> motif_starts = find_motif(a_motif);
781 for(std::vector<int>::iterator motif_start_i = motif_starts.begin();
782 motif_start_i != motif_starts.end();
785 motif_list.push_back(motif(*motif_start_i, a_motif));
789 void Sequence::clear_motifs()
794 const std::list<motif>& Sequence::motifs() const
800 Sequence::find_motif(std::string a_motif)
802 std::vector<int> motif_match_starts;
803 std::string a_motif_rc;
805 motif_match_starts.clear();
807 //std::cout << "motif is: " << a_motif << std::endl;
808 a_motif = motif_normalize(a_motif);
809 //std::cout << "motif is: " << a_motif << std::endl;
813 //std::cout << "Sequence: none blank motif\n";
814 motif_scan(a_motif, &motif_match_starts);
816 a_motif_rc = rc_motif(a_motif);
817 // make sure not to do search again if it is a palindrome
818 if (a_motif_rc != a_motif)
819 motif_scan(a_motif_rc, &motif_match_starts);
821 return motif_match_starts;
825 Sequence::motif_scan(std::string a_motif, std::vector<int> * motif_match_starts)
828 std::string::size_type seq_i;
829 int motif_i, motif_len;
831 // faster to loop thru the sequence as a old c std::string (ie char array)
833 //std::cout << "Sequence: motif, seq len = " << sequence.length() << std::endl;
834 motif_len = a_motif.length();
836 //std::cout << "motif_length: " << motif_len << std::endl;
837 //std::cout << "RAAARRRRR\n";
841 //std::cout << "motif: " << a_motif << std::endl;
843 //std::cout << "Sequence: motif, length= " << length << std::endl;
845 while (seq_i < length())
847 //std::cout << seq_c[seq_i];
848 //std::cout << seq_c[seq_i] << "?" << a_motif[motif_i] << ":" << motif_i << " ";
849 // this is pretty much a straight translation of Nora's python code
850 // to match iupac letter codes
851 if (a_motif[motif_i] =='N')
853 else if (a_motif[motif_i] == seq_c[seq_i])
855 else if ((a_motif[motif_i] =='M') &&
856 ((seq_c[seq_i]=='A') || (seq_c[seq_i]=='C')))
858 else if ((a_motif[motif_i] =='R') &&
859 ((seq_c[seq_i]=='A') || (seq_c[seq_i]=='G')))
861 else if ((a_motif[motif_i] =='W') &&
862 ((seq_c[seq_i]=='A') || (seq_c[seq_i]=='T')))
864 else if ((a_motif[motif_i] =='S') &&
865 ((seq_c[seq_i]=='C') || (seq_c[seq_i]=='G')))
867 else if ((a_motif[motif_i] =='Y') &&
868 ((seq_c[seq_i]=='C') || (seq_c[seq_i]=='T')))
870 else if ((a_motif[motif_i] =='K') &&
871 ((seq_c[seq_i]=='G') || (seq_c[seq_i]=='T')))
873 else if ((a_motif[motif_i] =='V') &&
874 ((seq_c[seq_i]=='A') || (seq_c[seq_i]=='C') ||
875 (seq_c[seq_i]=='G')))
877 else if ((a_motif[seq_i] =='H') &&
878 ((seq_c[seq_i]=='A') || (seq_c[seq_i]=='C') ||
879 (seq_c[seq_i]=='T')))
881 else if ((a_motif[motif_i] =='D') &&
882 ((seq_c[seq_i]=='A') || (seq_c[seq_i]=='G') ||
883 (seq_c[seq_i]=='T')))
885 else if ((a_motif[motif_i] =='B') &&
886 ((seq_c[seq_i]=='C') || (seq_c[seq_i]=='G') ||
887 (seq_c[seq_i]=='T')))
896 // end Nora stuff, now we see if a match is found this pass
897 if (motif_i == motif_len)
901 motif_match_starts->push_back(seq_i - motif_len + 1);
907 //std::cout << std::endl;
910 void Sequence::add_string_annotation(std::string a_seq,
913 std::vector<int> seq_starts = find_motif(a_seq);
915 //std::cout << "searching for " << a_seq << " found " << seq_starts.size() << std::endl;
917 for(std::vector<int>::iterator seq_start_i = seq_starts.begin();
918 seq_start_i != seq_starts.end();
921 annots.push_back(annot(*seq_start_i,
922 *seq_start_i+a_seq.size(),
928 void Sequence::find_sequences(std::list<Sequence>::iterator start,
929 std::list<Sequence>::iterator end)
931 while (start != end) {
932 add_string_annotation(*start, start->get_header());