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 <boost/filesystem/fstream.hpp>
16 namespace fs = boost::filesystem;
18 #include "alg/nway_paths.hpp"
19 #include "alg/conserved_path.hpp"
20 #include "mussa_exceptions.hpp"
27 NwayPaths::NwayPaths()
32 NwayPaths::setup(int w, int t)
35 soft_thres = threshold;
39 //cout << "nway: thres = " << threshold
40 // << ", soft threo = " << soft_thres << endl;
44 NwayPaths::set_soft_thres(int sft_thres)
46 soft_thres = sft_thres;
49 int NwayPaths::get_threshold() const
54 int NwayPaths::get_window() const
59 // dumbly goes thru and combines path windows exactly adjacent (ie + 1 index)
60 // doesn't deal with interleaved adjacency
62 NwayPaths::simple_refine()
64 // ext_path remembers the first window set in an extending path
65 ExtendedConservedPath ext_path, new_path;
66 list<ConservedPath>::iterator cur_path, next_path;
67 list<ConservedPath>::iterator pathz_i;
69 bool extending, end = false;
71 refined_pathz.clear();
73 //cout << "path number is: " << pathz.size() << endl;
74 pathz_i = pathz.begin();
76 // only try to extend when pathz isn't empty.
77 if (pathz_i != pathz.end())
79 ext_path = ExtendedConservedPath( win_size, *pathz_i);
81 while(pathz_i != pathz.end())
83 // keep track of current path and advance to next path
86 if (pathz_i == pathz.end())
93 // if node for each seq is equal to the next node+1 then for all
94 // sequences then we are extending
95 extending = cur_path->nextTo(*next_path);
106 // add the extend window length as first element and add as refined
107 // now that we have the path to extend save it
109 new_path.extend(win_ext_len);
110 refined_pathz.push_back(new_path);
113 ext_path = ExtendedConservedPath( win_size, *next_path);
117 //cout << "r_path number is: " << refined_pathz.size() << endl;
122 NwayPaths::add_path(int threshold, vector<int>& loaded_path)
124 pathz.push_back(ConservedPath(threshold, loaded_path));
128 NwayPaths::add_path(ConservedPath loaded_path)
130 pathz.push_back(loaded_path);
135 NwayPaths::save(fs::path save_file_path)
137 fs::fstream save_file;
138 list<ExtendedConservedPath >::iterator path_i, paths_end;
140 save_file.open(save_file_path, ios::out);
142 save_file << "<Mussa type=flp seq_count=" << sequence_count();
143 save_file << " win=" << win_size;
144 // add a function para new_thres defaults to -1 to later deal with
145 // reanalysis with higher thres - if statement whether to record base thres
146 // or new thres (ie if -1, then base)
147 save_file << " thres=" << threshold << " >\n";
149 path_i = refined_pathz.begin();
150 paths_end = refined_pathz.end();
151 //path_i = pathz.begin();
152 //paths_end = pathz.end();
153 while (path_i != paths_end)
155 ExtendedConservedPath& a_path = *path_i;
156 //cout << a_path.size() << endl;
157 //first entry is the window length of the windows in the path
158 save_file << a_path.window_size << ":";
159 for(size_t i = 0; i != sequence_count(); ++i)
161 save_file << a_path[i];
162 if (i != sequence_count())
169 save_file << "</Mussa>\n";
175 NwayPaths::sequence_count()
177 if (refined_pathz.begin() == refined_pathz.end() )
180 return refined_pathz.begin()->size();
185 NwayPaths::load(fs::path load_file_path)
187 fs::fstream load_file;
188 string file_data_line, header_data, data, path_node, path_width;
189 int space_split_i, equal_split_i, comma_split_i, colon_split_i;
190 vector<int> loaded_path;
192 load_file.open(load_file_path, ios::in);
196 throw mussa_load_error("Sequence File: " + load_file_path.string() + " not found");
201 // grab mussa tag - discard for now...maybe check in future...
202 getline(load_file,file_data_line);
203 space_split_i = file_data_line.find(" ");
204 file_data_line = file_data_line.substr(space_split_i+1);
205 // grab type tag - need in future to distinguish between flp and vlp paths
206 space_split_i = file_data_line.find(" ");
207 file_data_line = file_data_line.substr(space_split_i+1);
208 // get species/seq number
209 space_split_i = file_data_line.find(" ");
210 header_data = file_data_line.substr(0,space_split_i);
211 equal_split_i = header_data.find("=");
212 data = file_data_line.substr(equal_split_i+1);
213 unsigned int species_num = atoi (data.c_str());
214 file_data_line = file_data_line.substr(space_split_i+1);
216 space_split_i = file_data_line.find(" ");
217 header_data = file_data_line.substr(0,space_split_i);
218 equal_split_i = header_data.find("=");
219 data = file_data_line.substr(equal_split_i+1);
220 win_size = atoi (data.c_str());
221 file_data_line = file_data_line.substr(space_split_i+1);
222 // get threshold size
223 space_split_i = file_data_line.find(" ");
224 header_data = file_data_line.substr(0,space_split_i);
225 equal_split_i = header_data.find("=");
226 data = file_data_line.substr(equal_split_i+1);
227 threshold = atoi (data.c_str());
228 file_data_line = file_data_line.substr(space_split_i+1);
230 //cout << "seq_num=" << species_num << " win=" << win_size;
231 //cout << " thres=" << threshold << endl;
233 // clear out the current data
234 refined_pathz.clear();
238 getline(load_file,file_data_line);
239 while ( (!load_file.eof()) && (file_data_line != "</Mussa>") )
241 if (file_data_line != "")
244 colon_split_i = file_data_line.find(":");
245 // whats our window size?
246 path_width = file_data_line.substr(0,colon_split_i);
247 file_data_line = file_data_line.substr(colon_split_i+1);
248 for(size_t i = 0; i < species_num; i++)
250 comma_split_i = file_data_line.find(",");
251 path_node = file_data_line.substr(0, comma_split_i);
252 temp = atoi (path_node.c_str());
253 loaded_path.push_back(temp);
254 file_data_line = file_data_line.substr(comma_split_i+1);
256 assert (loaded_path.size() == species_num );
257 refined_pathz.push_back(ExtendedConservedPath(atoi(path_width.c_str()),
261 getline(load_file,file_data_line);
269 NwayPaths::path_search(vector<vector<FLPs> > all_comparisons, ConservedPath path, size_t depth)
271 list<int> new_nodes, trans_check_nodes;
272 list<int>::iterator new_nodes_i, new_nodes_end;
273 bool trans_check_good;
275 new_nodes = all_comparisons[depth - 1][depth].match_locations(path[depth-1]);
276 new_nodes_i = new_nodes.begin();
277 new_nodes_end = new_nodes.end();
278 while(new_nodes_i != new_nodes_end)
280 //cout << " * species " << depth << " node: " << *new_nodes_i << endl;
281 // check transitivity with previous nodes in path
282 trans_check_good = true;
283 for(size_t i = 0; i < depth - 1; i++)
285 trans_check_nodes = all_comparisons[i][depth].match_locations(path[i]);
286 if ( (trans_check_nodes.end() == find(trans_check_nodes.begin(),
287 trans_check_nodes.end(),
289 (trans_check_nodes.end() == find(trans_check_nodes.begin(),
290 trans_check_nodes.end(),
291 *new_nodes_i * -1) ) )
292 trans_check_good = false;
295 if (trans_check_good)
297 // this makes sure path nodes are recorded with RC status relative to
299 if ( path[depth-1] >= 0)
300 path.push_back(*new_nodes_i);
302 path.push_back(*new_nodes_i * -1);
304 if (depth < all_comparisons.size() - 1)
305 path_search(all_comparisons, path, depth + 1);
307 pathz.push_back(path);
313 /* use this if I ever get the friggin seqcomp match lists to sort...
314 if (binary_search(trans_check_nodes.begin(), trans_check_nodes.end(),
319 NwayPaths::find_paths_r(vector<vector<FLPs> > all_comparisons)
322 int win_i, window_num;
324 list<int>::iterator new_nodes_i, new_nodes_end;
327 window_num = all_comparisons[0][1].size();
328 // loop thru all windows in first species
329 for (win_i = 0; win_i < window_num; win_i++)
332 path.push_back(win_i);
333 new_nodes = all_comparisons[0][1].match_locations(path[0]);
334 new_nodes_i = new_nodes.begin();
335 new_nodes_end = new_nodes.end();
336 //if (new_nodes_i != new_nodes_end)
337 //cout << "* species 0 node: " << win_i << endl;
339 while(new_nodes_i != new_nodes_end)
341 //cout << " * species 1 node: " << *new_nodes_i << endl;
342 path[1] = *new_nodes_i;
343 path_search(all_comparisons, path, 2);
351 NwayPaths::save_old(fs::path save_file_path)
353 fs::fstream save_file;
354 list<ConservedPath >::iterator path_i, paths_end;
357 save_file.open(save_file_path, ios::app);
359 path_i = pathz.begin();
360 paths_end = pathz.end();
361 while(path_i != paths_end)
363 ConservedPath& a_path = *path_i;
364 //cout << a_path.size() << endl;
365 for(i = 0; i < path_i->size(); ++i)
366 save_file << i << "," << a_path[i] << " ";
376 NwayPaths::find_paths(vector<vector<FLPs> > all_comparisons)
379 vector<list<list<int> > > path_src_tree;
381 list<int>::iterator node_i, node_end;
382 <list<list<int> > >::iterator branch_i, branch_end;
385 path_src_tree.reserve(all_comparisons.size()- 1);
388 // loop thru all windows in first species
389 for (win_i = 0; win_i < window_num; win_i++)
391 // clear the path search tree
392 for(i = 0; i < all_comparisons.size(); i++)
393 path_src_tree[i].clear();
395 // top level kept empty even tho implicity has one entry of the first
396 // species at this window - why bother, adds a little speed
398 // get connection list for first species, creating a list of nodes
399 // of second species connected to the first species at this window
400 new_nodes = all_comparisons[0][1];
401 path_src_tree[1].push_back(new_nodes);
403 // loop thru rest of species for this window to see if any paths of matches
404 // go across all species
405 // if path search tree becomes empty, break out of loop, no reason to search further
407 while ((sp_i < all_comparisons.size()) && (path tree not empty))
409 branch_i = path_src_tree[1].begin();
410 branch_end = path_src_tree[1].end();
411 while (branch_i != branch_end)
413 node_i = branch_i->begin();
414 node_end = branch_i->end();
418 // loop over all current nodes
419 // get connection list for each node
420 // loop over each previous node in list
421 // get those nodes connection list
422 // intersect previous node connections with current
427 // insert any of the paths found into the master list of paths
429 // add no paths if tmp_pathz is empty...
433 void NwayPaths::refine()
439 void NwayPaths::print(list<vector<int> >& dump_path)
441 list<vector<int> >::iterator pathz_i;
442 vector<int>::iterator path_i;
444 cout << "printing list of lists\n";
445 for (pathz_i = dump_path.begin(); pathz_i != dump_path.end(); ++pathz_i)
447 for (path_i = pathz_i->begin(); path_i != pathz_i->end(); ++path_i)
448 cout << *path_i << " ";