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),
98 Sequence::Sequence(const char *seq)
102 set_filtered_sequence(seq);
105 Sequence::Sequence(const std::string& seq)
109 set_filtered_sequence(seq);
112 Sequence::Sequence(const Sequence& o)
117 motif_list(o.motif_list)
121 Sequence &Sequence::operator=(const Sequence& s)
124 //sequence = s.sequence;
129 motif_list = s.motif_list;
134 static void multiplatform_getline(std::istream& in, std::string& line)
139 while(in.good() and !(c == '\012' or c == '\015') ) {
143 // if we have cr-lf eat it
145 if (c=='\012' or c == '\015') {
150 //! load a fasta file into a sequence
152 * \param file_path the location of the fasta file in the filesystem
153 * \param seq_num which sequence in the file to load
154 * \param start_index starting position in the fasta sequence, 0 for beginning
155 * \param end_index ending position in the fasta sequence, 0 for end
156 * \return error message, empty string if no error. (gag!)
158 void Sequence::load_fasta(fs::path file_path, int seq_num,
159 int start_index, int end_index)
161 fs::fstream data_file;
162 data_file.open(file_path, std::ios::in);
164 if (!data_file.good())
166 throw mussa_load_error("Sequence File: "+file_path.string()+" not found");
169 load_fasta(data_file, seq_num, start_index, end_index);
170 } catch(sequence_empty_error e) {
171 // there doesn't appear to be any sequence
172 // catch and rethrow to include the filename
173 std::stringstream msg;
174 msg << "The selected sequence in "
175 << file_path.native_file_string()
176 << " appears to be empty";
177 throw sequence_empty_error(msg.str());
178 } catch(sequence_empty_file_error e) {
179 std::stringstream errormsg;
180 errormsg << file_path.native_file_string()
181 << " did not have any fasta sequences" << std::endl;
182 throw sequence_empty_file_error(errormsg.str());
188 Sequence::load_fasta(std::iostream& data_file, int seq_num,
189 int start_index, int end_index)
191 std::string file_data_line;
192 int header_counter = 0;
193 bool read_seq = true;
194 std::string rev_comp;
195 std::string sequence_raw;
196 std::string seq_tmp; // holds sequence during basic filtering
199 throw mussa_load_error("fasta sequence number is 1 based (can't be 0)");
202 // search for the header of the fasta sequence we want
203 while ( (!data_file.eof()) && (header_counter < seq_num) )
205 multiplatform_getline(data_file, file_data_line);
206 if (file_data_line.substr(0,1) == ">")
210 if (header_counter > 0) {
211 header = file_data_line.substr(1);
215 while ( !data_file.eof() && read_seq ) {
216 multiplatform_getline(data_file,file_data_line);
217 if (file_data_line.substr(0,1) == ">")
219 else sequence_raw += file_data_line;
222 // Lastly, if subselection of the sequence was specified we keep cut out
223 // and only keep that part
224 // end_index = 0 means no end was specified, so cut to the end
226 end_index = sequence_raw.size();
228 // sequence filtering for upcasing agctn and convert non AGCTN to N
229 if (end_index-start_index <= 0) {
230 std::string msg("The selected sequence appears to be empty");
231 throw sequence_empty_error(msg);
233 set_filtered_sequence(sequence_raw, start_index, end_index-start_index);
235 std::string errormsg("There were no fasta sequences");
236 throw sequence_empty_file_error(errormsg);
240 void Sequence::set_filtered_sequence(const std::string &old_seq,
241 std::string::size_type start,
242 std::string::size_type count)
244 char conversionTable[257];
247 count = old_seq.size() - start;
248 std::string::clear();
251 // Make a conversion table
253 // everything we don't specify below will become 'N'
254 for(int table_i=0; table_i < 256; table_i++)
256 conversionTable[table_i] = 'N';
258 // add end of string character for printing out table for testing purposes
259 conversionTable[256] = '\0';
261 // we want these to map to themselves - ie not to change
262 conversionTable[(int)'A'] = 'A';
263 conversionTable[(int)'T'] = 'T';
264 conversionTable[(int)'G'] = 'G';
265 conversionTable[(int)'C'] = 'C';
267 conversionTable[(int)'a'] = 'A';
268 conversionTable[(int)'t'] = 'T';
269 conversionTable[(int)'g'] = 'G';
270 conversionTable[(int)'c'] = 'C';
272 // finally, the actual conversion loop
273 for(std::string::size_type seq_index = 0; seq_index < count; seq_index++)
275 append(1, conversionTable[ (int)old_seq[seq_index+start]]);
279 // this doesn't work properly under gcc 3.x ... it can't recognize toupper
280 //transform(sequence.begin(), sequence.end(), sequence.begin(), toupper);
283 Sequence::load_annot(fs::path file_path, int start_index, int end_index)
285 fs::fstream data_stream(file_path, std::ios::in);
288 throw mussa_load_error("Sequence File: " + file_path.string() + " not found");
291 // so i should probably be passing the parse function some iterators
292 // but the annotations files are (currently) small, so i think i can
293 // get away with loading the whole file into memory
296 while(data_stream.good()) {
302 parse_annot(data, start_index, end_index);
305 /* If this works, yikes, this is some brain hurting code.
307 * what's going on is that when pb_annot is instantiated it stores references
308 * to begin, end, name, type, declared in the parse function, then
309 * when operator() is called it grabs values from those references
310 * and uses that to instantiate an annot object and append that to our
313 * This weirdness is because the spirit library requires that actions
314 * conform to a specific prototype operator()(IteratorT, IteratorT)
315 * which doesn't provide any useful opportunity for me to actually
316 * grab the results of our parsing.
318 * so I instantiate this structure in order to have a place to grab
322 struct push_back_annot {
323 std::list<annot>& annot_list;
329 push_back_annot(std::list<annot>& annot_list_,
334 : annot_list(annot_list_),
342 void operator()(std::string::const_iterator,
343 std::string::const_iterator) const
345 //std::cout << "adding annot: " << begin << "|" << end << "|" << name << "|" << type << std::endl;
346 annot_list.push_back(annot(begin, end, name, type));
350 struct push_back_seq {
351 std::list<Sequence>& seq_list;
355 push_back_seq(std::list<Sequence>& seq_list_,
358 : seq_list(seq_list_),
364 void operator()(std::string::const_iterator,
365 std::string::const_iterator) const
367 // filter out newlines from our sequence
369 for(std::string::const_iterator seq_i = seq.begin();
373 if (*seq_i != '\015' && *seq_i != '\012') new_seq += *seq_i;
375 //std::cout << "adding seq: " << name << " " << new_seq << std::endl;
379 seq_list.push_back(s);
384 Sequence::parse_annot(std::string data, int start_index, int end_index)
391 std::list<Sequence> query_seqs;
393 bool status = spirit::parse(data.begin(), data.end(),
400 )[spirit::assign_a(species)] >>
404 ( // parse an absolute location name
405 (spirit::uint_p[spirit::assign_a(start)] >>
407 spirit::uint_p[spirit::assign_a(end)] >>
412 )[spirit::assign_a(name)] >>
419 )[spirit::assign_a(type)]
421 // to understand how this group gets set
422 // read the comment above struct push_back_annot
423 )[push_back_annot(annots, start, end, type, name)]
425 (spirit::ch_p('>') >>
426 (*(spirit::print_p))[spirit::assign_a(name)] >>
428 (+(spirit::chset<>(iupac_alphabet)))[spirit::assign_a(seq)]
429 )[push_back_seq(query_seqs, name, seq)]
435 spirit::space_p*/).full;
437 // go seearch for query sequences
438 find_sequences(query_seqs.begin(), query_seqs.end());
441 const std::string& Sequence::get_species() const
446 void Sequence::add_annotation(const annot& a)
451 const std::list<annot>& Sequence::annotations() const
457 Sequence::subseq(int start, int count) const
459 // there might be an off by one error with start+count > size()
460 if ( count == npos || start+count > size()) {
461 count = size()-start;
463 Sequence new_seq(std::string::substr(start, count));
464 new_seq.set_header(get_header());
465 //new_seq.set_species(get_species());
467 new_seq.motif_list = motif_list;
468 // attempt to copy & reannotate position based annotations
469 int end = start+count;
471 for(std::list<annot>::const_iterator annot_i = annots.begin();
472 annot_i != annots.end();
475 int annot_start = annot_i->start;
476 int annot_end = annot_i->end;
478 if (annot_start < end) {
479 if (annot_start >=start) {
480 annot_start -= start;
485 if (annot_end < end) {
491 annot new_annot(annot_start, annot_end, annot_i->type, annot_i->name);
492 new_seq.annots.push_back(new_annot);
500 Sequence::rev_comp() const
502 std::string rev_comp;
503 char conversionTable[257];
504 int seq_i, table_i, len;
507 rev_comp.reserve(len);
508 // make a conversion table
509 // init all parts of conversion table to '~' character
510 // '~' I doubt will ever appear in a sequence file (jeez, I hope)
511 // and may the fleas of 1000 camels infest the genitals of any biologist (and
512 // seven generations of their progeny) who decides to make it mean
513 // something special!!!
514 // PS - double the curse for any smartass non-biologist who tries it as well
515 for(table_i=0; table_i < 256; table_i++)
517 conversionTable[table_i] = '~';
519 // add end of string character for printing out table for testing purposes
520 conversionTable[256] = '\0';
522 // add in the characters for the bases we want to convert
523 conversionTable[(int)'A'] = 'T';
524 conversionTable[(int)'T'] = 'A';
525 conversionTable[(int)'G'] = 'C';
526 conversionTable[(int)'C'] = 'G';
527 conversionTable[(int)'N'] = 'N';
529 // finally, the actual conversion loop
530 for(seq_i = len - 1; seq_i >= 0; seq_i--)
532 table_i = (int) at(seq_i);
533 rev_comp += conversionTable[table_i];
539 void Sequence::set_header(std::string header_)
545 Sequence::get_header() const
561 std::string::clear();
568 Sequence::save(fs::fstream &save_file)
569 //std::string save_file_path)
572 std::list<annot>::iterator annots_i;
574 // not sure why, or if i'm doing something wrong, but can't seem to pass
575 // file pointers down to this method from the mussa control class
576 // so each call to save a sequence appends to the file started by mussa_class
577 //save_file.open(save_file_path.c_str(), std::ios::app);
579 save_file << "<Sequence>" << std::endl;
580 save_file << *this << std::endl;
581 save_file << "</Sequence>" << std::endl;
583 save_file << "<Annotations>" << std::endl;
584 save_file << species << std::endl;
585 for (annots_i = annots.begin(); annots_i != annots.end(); ++annots_i)
587 save_file << annots_i->start << " " << annots_i->end << " " ;
588 save_file << annots_i->name << " " << annots_i->type << std::endl;
590 save_file << "</Annotations>" << std::endl;
595 Sequence::load_museq(fs::path load_file_path, int seq_num)
597 fs::fstream load_file;
598 std::string file_data_line;
601 std::string::size_type space_split_i;
602 std::string annot_value;
605 load_file.open(load_file_path, std::ios::in);
608 // search for the seq_num-th sequence
609 while ( (!load_file.eof()) && (seq_counter < seq_num) )
611 getline(load_file,file_data_line);
612 if (file_data_line == "<Sequence>")
615 getline(load_file, file_data_line);
616 assign(file_data_line);
617 getline(load_file, file_data_line);
618 getline(load_file, file_data_line);
619 if (file_data_line == "<Annotations>")
621 getline(load_file, file_data_line);
622 species = file_data_line;
623 while ( (!load_file.eof()) && (file_data_line != "</Annotations>") )
625 getline(load_file,file_data_line);
626 if ((file_data_line != "") && (file_data_line != "</Annotations>"))
628 // need to get 4 values...almost same code 4 times...
629 // get annot start index
630 space_split_i = file_data_line.find(" ");
631 annot_value = file_data_line.substr(0,space_split_i);
632 an_annot.start = atoi (annot_value.c_str());
633 file_data_line = file_data_line.substr(space_split_i+1);
634 // get annot end index
635 space_split_i = file_data_line.find(" ");
636 annot_value = file_data_line.substr(0,space_split_i);
637 an_annot.end = atoi (annot_value.c_str());
639 if (space_split_i == std::string::npos) // no entry for type or name
641 std::cout << "seq, annots - no type or name\n";
645 else // else get annot type
647 file_data_line = file_data_line.substr(space_split_i+1);
648 space_split_i = file_data_line.find(" ");
649 annot_value = file_data_line.substr(0,space_split_i);
650 an_annot.type = annot_value;
651 if (space_split_i == std::string::npos) // no entry for name
653 std::cout << "seq, annots - no name\n";
656 else // get annot name
658 file_data_line = file_data_line.substr(space_split_i+1);
659 space_split_i = file_data_line.find(" ");
660 annot_value = file_data_line.substr(0,space_split_i);
661 an_annot.type = annot_value;
664 annots.push_back(an_annot); // don't forget to actually add the annot
666 //std::cout << "seq, annots: " << an_annot.start << ", " << an_annot.end
667 // << "-->" << an_annot.type << "::" << an_annot.name << std::endl;
675 Sequence::rc_motif(std::string a_motif)
677 std::string rev_comp;
678 char conversionTable[257];
679 int seq_i, table_i, len;
681 len = a_motif.length();
682 rev_comp.reserve(len);
684 for(table_i=0; table_i < 256; table_i++)
686 conversionTable[table_i] = '~';
688 // add end of std::string character for printing out table for testing purposes
689 conversionTable[256] = '\0';
691 // add in the characters for the bases we want to convert (IUPAC)
692 conversionTable[(int)'A'] = 'T';
693 conversionTable[(int)'T'] = 'A';
694 conversionTable[(int)'G'] = 'C';
695 conversionTable[(int)'C'] = 'G';
696 conversionTable[(int)'N'] = 'N';
697 conversionTable[(int)'M'] = 'K';
698 conversionTable[(int)'R'] = 'Y';
699 conversionTable[(int)'W'] = 'W';
700 conversionTable[(int)'S'] = 'S';
701 conversionTable[(int)'Y'] = 'R';
702 conversionTable[(int)'K'] = 'M';
703 conversionTable[(int)'V'] = 'B';
704 conversionTable[(int)'H'] = 'D';
705 conversionTable[(int)'D'] = 'H';
706 conversionTable[(int)'B'] = 'V';
708 // finally, the actual conversion loop
709 for(seq_i = len - 1; seq_i >= 0; seq_i--)
711 //std::cout << "** i = " << seq_i << " bp = " <<
712 table_i = (int) a_motif[seq_i];
713 rev_comp += conversionTable[table_i];
716 //std::cout << "seq: " << a_motif << std::endl;
717 //std::cout << "rc: " << rev_comp << std::endl;
723 Sequence::motif_normalize(std::string a_motif)
725 std::string valid_motif;
728 len = a_motif.length();
729 valid_motif.reserve(len);
731 // this just upcases IUPAC symbols. Eventually should return an error if non IUPAC is present.
732 // current nonIUPAC symbols are omitted, which is not reported atm
733 for(seq_i = 0; seq_i < len; seq_i++)
735 if ((a_motif[seq_i] == 'a') || (a_motif[seq_i] == 'A'))
737 else if ((a_motif[seq_i] == 't') || (a_motif[seq_i] == 'T'))
739 else if ((a_motif[seq_i] == 'g') || (a_motif[seq_i] == 'G'))
741 else if ((a_motif[seq_i] == 'c') || (a_motif[seq_i] == 'C'))
743 else if ((a_motif[seq_i] == 'n') || (a_motif[seq_i] == 'N'))
745 else if ((a_motif[seq_i] == 'm') || (a_motif[seq_i] == 'M'))
747 else if ((a_motif[seq_i] == 'r') || (a_motif[seq_i] == 'R'))
749 else if ((a_motif[seq_i] == 'w') || (a_motif[seq_i] == 'W'))
751 else if ((a_motif[seq_i] == 's') || (a_motif[seq_i] == 'S'))
753 else if ((a_motif[seq_i] == 'y') || (a_motif[seq_i] == 'Y'))
755 else if ((a_motif[seq_i] == 'k') || (a_motif[seq_i] == 'K'))
757 else if ((a_motif[seq_i] == 'v') || (a_motif[seq_i] == 'V'))
759 else if ((a_motif[seq_i] == 'h') || (a_motif[seq_i] == 'H'))
761 else if ((a_motif[seq_i] == 'd') || (a_motif[seq_i] == 'D'))
763 else if ((a_motif[seq_i] == 'b') || (a_motif[seq_i] == 'B'))
766 std::string msg = "Letter ";
767 msg += a_motif[seq_i];
768 msg += " is not a valid IUPAC symbol";
769 throw motif_normalize_error(msg);
772 //std::cout << "valid_motif is: " << valid_motif << std::endl;
776 void Sequence::add_motif(std::string a_motif)
778 std::vector<int> motif_starts = find_motif(a_motif);
780 for(std::vector<int>::iterator motif_start_i = motif_starts.begin();
781 motif_start_i != motif_starts.end();
784 motif_list.push_back(motif(*motif_start_i, a_motif));
788 void Sequence::clear_motifs()
793 const std::list<motif>& Sequence::motifs() const
799 Sequence::find_motif(std::string a_motif)
801 std::vector<int> motif_match_starts;
802 std::string a_motif_rc;
804 motif_match_starts.clear();
806 //std::cout << "motif is: " << a_motif << std::endl;
807 a_motif = motif_normalize(a_motif);
808 //std::cout << "motif is: " << a_motif << std::endl;
812 //std::cout << "Sequence: none blank motif\n";
813 motif_scan(a_motif, &motif_match_starts);
815 a_motif_rc = rc_motif(a_motif);
816 // make sure not to do search again if it is a palindrome
817 if (a_motif_rc != a_motif)
818 motif_scan(a_motif_rc, &motif_match_starts);
820 return motif_match_starts;
824 Sequence::motif_scan(std::string a_motif, std::vector<int> * motif_match_starts)
827 std::string::size_type seq_i;
828 int motif_i, motif_len;
830 // faster to loop thru the sequence as a old c std::string (ie char array)
832 //std::cout << "Sequence: motif, seq len = " << sequence.length() << std::endl;
833 motif_len = a_motif.length();
835 //std::cout << "motif_length: " << motif_len << std::endl;
836 //std::cout << "RAAARRRRR\n";
840 //std::cout << "motif: " << a_motif << std::endl;
842 //std::cout << "Sequence: motif, length= " << length << std::endl;
844 while (seq_i < length())
846 //std::cout << seq_c[seq_i];
847 //std::cout << seq_c[seq_i] << "?" << a_motif[motif_i] << ":" << motif_i << " ";
848 // this is pretty much a straight translation of Nora's python code
849 // to match iupac letter codes
850 if (a_motif[motif_i] =='N')
852 else if (a_motif[motif_i] == seq_c[seq_i])
854 else if ((a_motif[motif_i] =='M') &&
855 ((seq_c[seq_i]=='A') || (seq_c[seq_i]=='C')))
857 else if ((a_motif[motif_i] =='R') &&
858 ((seq_c[seq_i]=='A') || (seq_c[seq_i]=='G')))
860 else if ((a_motif[motif_i] =='W') &&
861 ((seq_c[seq_i]=='A') || (seq_c[seq_i]=='T')))
863 else if ((a_motif[motif_i] =='S') &&
864 ((seq_c[seq_i]=='C') || (seq_c[seq_i]=='G')))
866 else if ((a_motif[motif_i] =='Y') &&
867 ((seq_c[seq_i]=='C') || (seq_c[seq_i]=='T')))
869 else if ((a_motif[motif_i] =='K') &&
870 ((seq_c[seq_i]=='G') || (seq_c[seq_i]=='T')))
872 else if ((a_motif[motif_i] =='V') &&
873 ((seq_c[seq_i]=='A') || (seq_c[seq_i]=='C') ||
874 (seq_c[seq_i]=='G')))
876 else if ((a_motif[seq_i] =='H') &&
877 ((seq_c[seq_i]=='A') || (seq_c[seq_i]=='C') ||
878 (seq_c[seq_i]=='T')))
880 else if ((a_motif[motif_i] =='D') &&
881 ((seq_c[seq_i]=='A') || (seq_c[seq_i]=='G') ||
882 (seq_c[seq_i]=='T')))
884 else if ((a_motif[motif_i] =='B') &&
885 ((seq_c[seq_i]=='C') || (seq_c[seq_i]=='G') ||
886 (seq_c[seq_i]=='T')))
895 // end Nora stuff, now we see if a match is found this pass
896 if (motif_i == motif_len)
900 motif_match_starts->push_back(seq_i - motif_len + 1);
906 //std::cout << std::endl;
909 void Sequence::add_string_annotation(std::string a_seq,
912 std::vector<int> seq_starts = find_motif(a_seq);
914 //std::cout << "searching for " << a_seq << " found " << seq_starts.size() << std::endl;
916 for(std::vector<int>::iterator seq_start_i = seq_starts.begin();
917 seq_start_i != seq_starts.end();
920 annots.push_back(annot(*seq_start_i,
921 *seq_start_i+a_seq.size(),
927 void Sequence::find_sequences(std::list<Sequence>::iterator start,
928 std::list<Sequence>::iterator end)
930 while (start != end) {
931 add_string_annotation(*start, start->get_header());