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_nway.cc -----------
13 // ----------------------------------------
15 #include "alg/nway_paths.hpp"
16 #include "alg/conserved_path.hpp"
17 #include "mussa_exceptions.hpp"
25 NwayPaths::NwayPaths()
30 NwayPaths::setup(int w, int t)
33 soft_thres = threshold;
37 //cout << "nway: thres = " << threshold
38 // << ", soft threo = " << soft_thres << endl;
42 NwayPaths::set_soft_thres(int sft_thres)
44 soft_thres = sft_thres;
47 int NwayPaths::get_threshold() const
52 int NwayPaths::get_window() const
57 // dumbly goes thru and combines path windows exactly adjacent (ie + 1 index)
58 // doesn't deal with interleaved adjacency
60 NwayPaths::simple_refine()
62 // ext_path remembers the first window set in an extending path
63 ExtendedConservedPath ext_path, new_path;
64 list<ConservedPath>::iterator cur_path, next_path;
65 list<ConservedPath>::iterator pathz_i;
67 bool extending, end = false;
69 refined_pathz.clear();
71 //cout << "path number is: " << pathz.size() << endl;
72 pathz_i = pathz.begin();
74 // only try to extend when pathz isn't empty.
75 if (pathz_i != pathz.end())
77 ext_path = ExtendedConservedPath( win_size, *pathz_i);
79 while(pathz_i != pathz.end())
81 // keep track of current path and advance to next path
84 if (pathz_i == pathz.end())
91 // if node for each seq is equal to the next node+1 then for all
92 // sequences then we are extending
93 extending = cur_path->nextTo(*next_path);
104 // add the extend window length as first element and add as refined
105 // now that we have the path to extend save it
107 new_path.extend(win_ext_len);
108 refined_pathz.push_back(new_path);
111 ext_path = ExtendedConservedPath( win_size, *next_path);
115 //cout << "r_path number is: " << refined_pathz.size() << endl;
120 NwayPaths::add_path(int threshold, vector<int>& loaded_path)
122 pathz.push_back(ConservedPath(threshold, loaded_path));
126 NwayPaths::add_path(ConservedPath loaded_path)
128 pathz.push_back(loaded_path);
133 NwayPaths::save(string save_file_path)
136 list<ExtendedConservedPath >::iterator path_i, paths_end;
138 save_file.open(save_file_path.c_str(), ios::out);
140 save_file << "<Mussa type=flp seq_count=" << sequence_count();
141 save_file << " win=" << win_size;
142 // add a function para new_thres defaults to -1 to later deal with
143 // reanalysis with higher thres - if statement whether to record base thres
144 // or new thres (ie if -1, then base)
145 save_file << " thres=" << threshold << " >\n";
147 path_i = refined_pathz.begin();
148 paths_end = refined_pathz.end();
149 //path_i = pathz.begin();
150 //paths_end = pathz.end();
151 while (path_i != paths_end)
153 ExtendedConservedPath& a_path = *path_i;
154 //cout << a_path.size() << endl;
155 //first entry is the window length of the windows in the path
156 save_file << a_path.window_size << ":";
157 for(size_t i = 0; i != sequence_count(); ++i)
159 save_file << a_path[i];
160 if (i != sequence_count())
167 save_file << "</Mussa>\n";
173 NwayPaths::sequence_count()
175 if (refined_pathz.begin() == refined_pathz.end() )
178 return refined_pathz.begin()->size();
183 NwayPaths::load(string load_file_path)
186 string file_data_line, header_data, data, path_node, path_width;
187 int space_split_i, equal_split_i, comma_split_i, colon_split_i;
188 vector<int> loaded_path;
190 load_file.open(load_file_path.c_str(), ios::in);
194 throw mussa_load_error("Sequence File: " + load_file_path + " not found");
199 // grab mussa tag - discard for now...maybe check in future...
200 getline(load_file,file_data_line);
201 space_split_i = file_data_line.find(" ");
202 file_data_line = file_data_line.substr(space_split_i+1);
203 // grab type tag - need in future to distinguish between flp and vlp paths
204 space_split_i = file_data_line.find(" ");
205 file_data_line = file_data_line.substr(space_split_i+1);
206 // get species/seq number
207 space_split_i = file_data_line.find(" ");
208 header_data = file_data_line.substr(0,space_split_i);
209 equal_split_i = header_data.find("=");
210 data = file_data_line.substr(equal_split_i+1);
211 unsigned int species_num = atoi (data.c_str());
212 file_data_line = file_data_line.substr(space_split_i+1);
214 space_split_i = file_data_line.find(" ");
215 header_data = file_data_line.substr(0,space_split_i);
216 equal_split_i = header_data.find("=");
217 data = file_data_line.substr(equal_split_i+1);
218 win_size = atoi (data.c_str());
219 file_data_line = file_data_line.substr(space_split_i+1);
220 // get threshold size
221 space_split_i = file_data_line.find(" ");
222 header_data = file_data_line.substr(0,space_split_i);
223 equal_split_i = header_data.find("=");
224 data = file_data_line.substr(equal_split_i+1);
225 threshold = atoi (data.c_str());
226 file_data_line = file_data_line.substr(space_split_i+1);
228 //cout << "seq_num=" << species_num << " win=" << win_size;
229 //cout << " thres=" << threshold << endl;
231 // clear out the current data
232 refined_pathz.clear();
236 getline(load_file,file_data_line);
237 while ( (!load_file.eof()) && (file_data_line != "</Mussa>") )
239 if (file_data_line != "")
242 colon_split_i = file_data_line.find(":");
243 // whats our window size?
244 path_width = file_data_line.substr(0,colon_split_i);
245 file_data_line = file_data_line.substr(colon_split_i+1);
246 for(size_t i = 0; i < species_num; i++)
248 comma_split_i = file_data_line.find(",");
249 path_node = file_data_line.substr(0, comma_split_i);
250 temp = atoi (path_node.c_str());
251 loaded_path.push_back(temp);
252 file_data_line = file_data_line.substr(comma_split_i+1);
254 assert (loaded_path.size() == species_num );
255 refined_pathz.push_back(ExtendedConservedPath(atoi(path_width.c_str()),
259 getline(load_file,file_data_line);
267 NwayPaths::path_search(vector<vector<FLPs> > all_comparisons, ConservedPath path, size_t depth)
269 list<int> new_nodes, trans_check_nodes;
270 list<int>::iterator new_nodes_i, new_nodes_end;
271 bool trans_check_good;
273 new_nodes = all_comparisons[depth - 1][depth].match_locations(path[depth-1]);
274 new_nodes_i = new_nodes.begin();
275 new_nodes_end = new_nodes.end();
276 while(new_nodes_i != new_nodes_end)
278 //cout << " * species " << depth << " node: " << *new_nodes_i << endl;
279 // check transitivity with previous nodes in path
280 trans_check_good = true;
281 for(size_t i = 0; i < depth - 1; i++)
283 trans_check_nodes = all_comparisons[i][depth].match_locations(path[i]);
284 if ( (trans_check_nodes.end() == find(trans_check_nodes.begin(),
285 trans_check_nodes.end(),
287 (trans_check_nodes.end() == find(trans_check_nodes.begin(),
288 trans_check_nodes.end(),
289 *new_nodes_i * -1) ) )
290 trans_check_good = false;
293 if (trans_check_good)
295 // this makes sure path nodes are recorded with RC status relative to
297 if ( path[depth-1] >= 0)
298 path.push_back(*new_nodes_i);
300 path.push_back(*new_nodes_i * -1);
302 if (depth < all_comparisons.size() - 1)
303 path_search(all_comparisons, path, depth + 1);
305 pathz.push_back(path);
311 /* use this if I ever get the friggin seqcomp match lists to sort...
312 if (binary_search(trans_check_nodes.begin(), trans_check_nodes.end(),
317 NwayPaths::find_paths_r(vector<vector<FLPs> > all_comparisons)
320 int win_i, window_num;
322 list<int>::iterator new_nodes_i, new_nodes_end;
325 window_num = all_comparisons[0][1].size();
326 // loop thru all windows in first species
327 for (win_i = 0; win_i < window_num; win_i++)
330 path.push_back(win_i);
331 new_nodes = all_comparisons[0][1].match_locations(path[0]);
332 new_nodes_i = new_nodes.begin();
333 new_nodes_end = new_nodes.end();
334 //if (new_nodes_i != new_nodes_end)
335 //cout << "* species 0 node: " << win_i << endl;
337 while(new_nodes_i != new_nodes_end)
339 //cout << " * species 1 node: " << *new_nodes_i << endl;
340 path[1] = *new_nodes_i;
341 path_search(all_comparisons, path, 2);
349 NwayPaths::save_old(string save_file_path)
352 list<ConservedPath >::iterator path_i, paths_end;
355 save_file.open(save_file_path.c_str(), ios::app);
357 path_i = pathz.begin();
358 paths_end = pathz.end();
359 while(path_i != paths_end)
361 ConservedPath& a_path = *path_i;
362 //cout << a_path.size() << endl;
363 for(i = 0; i < path_i->size(); ++i)
364 save_file << i << "," << a_path[i] << " ";
374 NwayPaths::find_paths(vector<vector<FLPs> > all_comparisons)
377 vector<list<list<int> > > path_src_tree;
379 list<int>::iterator node_i, node_end;
380 <list<list<int> > >::iterator branch_i, branch_end;
383 path_src_tree.reserve(all_comparisons.size()- 1);
386 // loop thru all windows in first species
387 for (win_i = 0; win_i < window_num; win_i++)
389 // clear the path search tree
390 for(i = 0; i < all_comparisons.size(); i++)
391 path_src_tree[i].clear();
393 // top level kept empty even tho implicity has one entry of the first
394 // species at this window - why bother, adds a little speed
396 // get connection list for first species, creating a list of nodes
397 // of second species connected to the first species at this window
398 new_nodes = all_comparisons[0][1];
399 path_src_tree[1].push_back(new_nodes);
401 // loop thru rest of species for this window to see if any paths of matches
402 // go across all species
403 // if path search tree becomes empty, break out of loop, no reason to search further
405 while ((sp_i < all_comparisons.size()) && (path tree not empty))
407 branch_i = path_src_tree[1].begin();
408 branch_end = path_src_tree[1].end();
409 while (branch_i != branch_end)
411 node_i = branch_i->begin();
412 node_end = branch_i->end();
416 // loop over all current nodes
417 // get connection list for each node
418 // loop over each previous node in list
419 // get those nodes connection list
420 // intersect previous node connections with current
425 // insert any of the paths found into the master list of paths
427 // add no paths if tmp_pathz is empty...
431 void NwayPaths::refine()
437 void NwayPaths::print(list<vector<int> >& dump_path)
439 list<vector<int> >::iterator pathz_i;
440 vector<int>::iterator path_i;
442 cout << "printing list of lists\n";
443 for (pathz_i = dump_path.begin(); pathz_i != dump_path.end(); ++pathz_i)
445 for (path_i = pathz_i->begin(); path_i != pathz_i->end(); ++path_i)
446 cout << *path_i << " ";