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 #include <QString>
  18
  19 #include <boost/filesystem/path.hpp>
  20
  21 #include <list>
  22 #include <string>
  23 #include <vector>
  24
  25 #include "alg/conserved_path.hpp"
  26 #include "alg/flp.hpp"
  27 #include "alg/sequence.hpp"
  28
  29 class NwayPaths : public QObject
  30 {
  31   Q_OBJECT
  32
  33 signals:
  34   //! emit to indicate how much progress we've made
  35   void progress(const QString& description, int cur, int max);
  36
  37 public:
  38     typedef std::list<ConservedPath> ConservedPaths;
  39
  40     NwayPaths();
  41     NwayPaths(const NwayPaths&);
  42
  43     //! setup an nway comparison, initialize # of species, window size,
  44     //! threshold
  45     void setup(int w, int t);
  46     void setup_ent(double new_entropy_thres, std::vector<Sequence> some_Seqs);
  47     //! clear out our path
  48     void clear();
  49     //! get the "soft" threshold (between the hard threshold and window size)
  50     int get_soft_threshold() const;
  51     //! set the score that a match must exceed inorder to be recorded as a path
  52     void set_soft_threshold(int soft_thres);
  53     //! return minimum threshold for this analysis
  54     int get_threshold() const;
  55     //! return window size used for this analysis
  56     int get_window() const;
  57
  58     void radiate_path_search(std::vector<std::vector<FLPs> > all_comparisons);
  59     void trans_path_search(std::vector<std::vector<FLPs> > all_comparisons);
  60     void entropy_path_search(std::vector<std::vector<FLPs> > all_comparisons);
  61     double path_entropy(std::vector<int> path);
  62
  63   // old recursive transitive nway ... has issues checking all links?
  64     void find_paths_r(std::vector<std::vector<FLPs> > all_comparisons);
  65     void path_search(std::vector<std::vector<FLPs> > all_comparisons, ConservedPath path, size_t depth);
  66
  67     void simple_refine();
  68     void save(boost::filesystem::path save_file_path);
  69     //! load a muway file, \throws mussa_load_error
  70     void load(boost::filesystem::path load_file_path);
  71     void add_path(int threshold, std::vector<int>& loaded_path);
  72     void add_path(ConservedPath loaded_path);
  73     //! how many sequences are in our comparison
  74     size_t sequence_count();
  75
  76     void find_paths(std::vector<std::vector<FLPs> > all_comparisons);
  77     void refine();
  78
  79     void save_old(boost::filesystem::path save_file_path);
  80     void print(std::list<std::vector<int> >&);
  81
  82     // The following iterator functions are mostly for the python interface
  83     // they'll have problems when being called from within a const object
  84     ConservedPaths::iterator pbegin() { return pathz.begin() ; }
  85     ConservedPaths::iterator pend() { return pathz.end() ; }
  86     size_t path_size() const { return refined_pathz.size(); }
  87     ConservedPaths::iterator rpbegin() { return refined_pathz.begin() ; }
  88     ConservedPaths::iterator rpend() { return refined_pathz.end() ; }
  89     size_t refined_path_size() const { return refined_pathz.size(); }
  90
  91     // these probably shouldn't be public, but lets start
  92     // simple
  93     ConservedPaths pathz;
  94     ConservedPaths refined_pathz;
  95
  96 protected:
  97     int threshold;
  98     size_t win_size;
  99     int soft_thres;
 100
 101     double ent_thres;
 102     std::vector<char *> c_sequences; //used by entropy_path_search
 103
 104 };
 105 #endif