alg/nway_paths.hpp

   1 #ifndef _MUSSA_NWAY_H_
   2 #define _MUSSA_NWAY_H_
   3 //  This file is part of the Mussa source distribution.
   4 //  http://mussa.caltech.edu/
   5 //  Contact author: Tristan  De Buysscher, tristan@caltech.edu
   6
   7 // This program and all associated source code files are Copyright (C) 2005
   8 // the California Institute of Technology, Pasadena, CA, 91125 USA.  It is
   9 // under the GNU Public License; please see the included LICENSE.txt
  10 // file for more information, or contact Tristan directly.
  11
  12
  13 //                        ----------------------------------------
  14 //                         ----------  mussa_nway.hh  -----------
  15 //                        ----------------------------------------
  16 #include <QObject>
  17
  18 #include <boost/filesystem/path.hpp>
  19
  20 #include <list>
  21 #include <string>
  22 #include <vector>
  23
  24 #include "alg/conserved_path.hpp"
  25 #include "alg/flp.hpp"
  26 #include "alg/sequence.hpp"
  27
  28 class NwayPaths : public QObject
  29 {
  30   Q_OBJECT
  31
  32 signals:
  33   //! emit to indicate how much progress we've made
  34   void progress(const std::string& description, int cur, int max);
  35
  36 public:
  37     NwayPaths();
  38     NwayPaths(const NwayPaths&);
  39
  40     //! setup an nway comparison, initialize # of species, window size,
  41     //! threshold
  42     void setup(int w, int t);
  43     void setup_ent(double new_entropy_thres, std::vector<Sequence> some_Seqs);
  44     //! clear out our path
  45     void clear();
  46     //! set the score that a match must exceed inorder to be recorded as a path
  47     void set_soft_threshold(int soft_thres);
  48     //! return minimum threshold for this analysis
  49     int get_threshold() const;
  50     //! return window size used for this analysis
  51     int get_window() const;
  52
  53     void radiate_path_search(std::vector<std::vector<FLPs> > all_comparisons);
  54     void trans_path_search(std::vector<std::vector<FLPs> > all_comparisons);
  55     void entropy_path_search(std::vector<std::vector<FLPs> > all_comparisons);
  56     double path_entropy(std::vector<int> path);
  57
  58   // old recursive transitive nway ... has issues checking all links?
  59     void find_paths_r(std::vector<std::vector<FLPs> > all_comparisons);
  60     void path_search(std::vector<std::vector<FLPs> > all_comparisons, ConservedPath path, size_t depth);
  61
  62     void simple_refine();
  63     void save(boost::filesystem::path save_file_path);
  64     //! load a muway file, \throws mussa_load_error
  65     void load(boost::filesystem::path load_file_path);
  66     void add_path(int threshold, std::vector<int>& loaded_path);
  67     void add_path(ConservedPath loaded_path);
  68     //! how many sequences are in our comparison
  69     size_t sequence_count();
  70
  71     void find_paths(std::vector<std::vector<FLPs> > all_comparisons);
  72     void refine();
  73
  74     void save_old(boost::filesystem::path save_file_path);
  75     void print(std::list<std::vector<int> >&);
  76
  77     // The following iterator functions are mostly for the python interface
  78     // they'll have problems when being called from within a const object
  79     std::list<ConservedPath>::iterator pbegin() { return pathz.begin() ; }
  80     std::list<ConservedPath>::iterator pend() { return pathz.end() ; }
  81     size_t path_size() const { return refined_pathz.size(); }
  82     std::list<ConservedPath>::iterator rpbegin() { return refined_pathz.begin() ; }
  83     std::list<ConservedPath>::iterator rpend() { return refined_pathz.end() ; }
  84     size_t refined_path_size() const { return refined_pathz.size(); }
  85
  86     // these probably shouldn't be public, but lets start
  87     // simple
  88     std::list<ConservedPath> pathz;
  89     std::list<ConservedPath > refined_pathz;
  90
  91 protected:
  92     int threshold;
  93     size_t win_size;
  94     int soft_thres;
  95
  96     double ent_thres;
  97     std::vector<char *> c_sequences; //used by entropy_path_search
  98
  99 };
 100 #endif