src/genealgorithms.cpp

   1 /***************************************************************************
   2  *   Copyright (C) 2006 by Michael Andreen                                 *
   3  *   andreen@student.chalmers.se                                           *
   4  *                                                                         *
   5  *   This program is free software; you can redistribute it and/or modify  *
   6  *   it under the terms of the GNU General Public License as published by  *
   7  *   the Free Software Foundation; either version 2 of the License, or     *
   8  *   (at your option) any later version.                                   *
   9  *                                                                         *
  10  *   This program is distributed in the hope that it will be useful,       *
  11  *   but WITHOUT ANY WARRANTY; without even the implied warranty of        *
  12  *   MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the         *
  13  *   GNU General Public License for more details.                          *
  14  *                                                                         *
  15  *   You should have received a copy of the GNU General Public License     *
  16  *   along with this program; if not, write to the                         *
  17  *   Free Software Foundation, Inc.,                                       *
  18  *   51 Franklin Street, Fifth Floor, Boston, MA  02110-1301, USA          *
  19  ***************************************************************************/
  20
  21 #include "genealgorithms.h"
  22 #include "geneorder.h"
  23
  24 #include <algorithm>
  25 #include <set>
  26 #include <stack>
  27 #include <cstdlib>
  28 #include <iostream>
  29 using namespace std;
  30
  31 std::pair<int,int> longestSequences(const GeneOrder& go){
  32         vector<vector<int> > v = robinsonSchensted(go);
  33         return pair<int,int>(v[0].size(),v.size());
  34 }
  35
  36 std::vector<std::vector<int> > robinsonSchensted(const GeneOrder& go){
  37         vector<vector<int> > v;
  38         for (GeneOrder::iterator i = go.begin(); i != go.end(); ++i){
  39                 int n = abs(*i);
  40                 bool added = false;
  41                 for (vector<vector<int> >::iterator vs = v.begin();
  42                                 vs != v.end(); ++vs){
  43                         vector<int>::iterator bigger = upper_bound(vs->begin(),vs->end(),n);
  44                         if ( bigger == vs->end()){
  45                                 vs->push_back(n);
  46                                 added = true;
  47                                 break;
  48                         }else{
  49                                 swap(n,*bigger);
  50                         }
  51                 }
  52                 if (!added){
  53                         v.push_back(vector<int>());
  54                         v.back().push_back(n);
  55                 }
  56         }
  57         return v;
  58 }
  59
  60 struct FindP{
  61         size_t p;
  62         FindP(size_t p) : p(p) {}
  63         bool operator()(Interval i){
  64                 return (i.first == p || i.second == p);
  65         }
  66 };
  67
  68
  69 int countCycles(const GeneOrder& go){
  70         int cycles = 0;
  71         set<size_t> marked;
  72         vector<Interval> intervals = findIntervals(go);
  73         vector<Interval> points = findIntervalsAtPoints(intervals);
  74         for (size_t p = 1; p < go.size(); ++p){
  75                 if (marked.find(p) != marked.end())
  76                         continue;
  77                 Interval i = intervals[points[p].first];
  78                 while (marked.find(p) == marked.end()){
  79                         marked.insert(p);
  80                         if (i == intervals[points[p].first])
  81                                 i = intervals[points[p].second];
  82                         else
  83                                 i = intervals[points[p].first];
  84
  85                         if (p == i.first)
  86                                 p = i.second;
  87                         else
  88                                 p = i.first;
  89                 }
  90                 ++cycles;
  91         }
  92         return cycles;
  93 }
  94
  95 int sign(Gene g){
  96         if (g > 0)
  97                 return 1;
  98         if (g < 0)
  99                 return -1;
 100         return 0;
 101 }
 102
 103 struct Abs{
 104         Gene operator()(Gene x) const{
 105                 return abs(x);
 106         }
 107 };
 108 std::vector<Component> findComponents(const GeneOrder& go){
 109         vector<Component> components;
 110         vector<int> os(go.size()-1);
 111         for (size_t i = 0; i < os.size(); ++i)
 112                 os[i] = (go[i]*go[i+1] > 0 ? sign(go[i]) : 0);
 113         stack<Gene> Mdir;
 114         Mdir.push(go.size()-1);
 115         stack<Gene> Mrev;
 116         Mrev.push(0);
 117         stack<size_t> Sdir;
 118         Sdir.push(0);
 119         stack<size_t> Srev;
 120         Srev.push(0);
 121         vector<Gene> dir;
 122         dir.push_back(go.size()-1);
 123         vector<Gene> rev;
 124         rev.push_back(0);
 125         size_t s;
 126         vector<Gene> p(go.list());
 127         transform(p.begin(),p.end(),p.begin(),Abs());
 128         for (size_t i = 1; i < go.size(); ++i){
 129                 //Directed
 130                 if (p[i-1] > p[i])
 131                         Mdir.push(p[i-1]);
 132                 else while (Mdir.top() < p[i])
 133                         Mdir.pop();
 134                 dir.push_back(Mdir.top());
 135
 136                 s = Sdir.top();
 137                 while(p[Sdir.top()] > p[i] || dir[Sdir.top()] < p[i]){
 138                         Sdir.pop();
 139                         os[Sdir.top()] = (os[Sdir.top()] == os[s] ? os[s] : 0);
 140                         s = Sdir.top();
 141                 }
 142                 if (go[i] > 0 && dir[i] == dir[s] && static_cast<Gene>(i - s) == p[i] - p[s])
 143                         components.push_back(Component(p[s],p[i],(s+1 == i ? 0 : os[s])));
 144
 145                 //Reverse
 146                 if (p[i-1] < p[i])
 147                         Mrev.push(p[i-1]);
 148                 else while (Mrev.top() > p[i])
 149                         Mrev.pop();
 150                 rev.push_back(Mrev.top());
 151
 152                 s = Srev.top();
 153                 while((p[s] < p[i] || rev[s] > p[i]) && s > 0){
 154                         Srev.pop();
 155                         os[Srev.top()] *= (os[Srev.top()] == os[s] ? 1 : 0);
 156                         s = Srev.top();
 157                 }
 158                 if (go[i] < 0 && rev[i] == rev[s] && static_cast<Gene>(i - s) == p[s] - p[i])
 159                         components.push_back(Component(-p[s],-p[i],(s+1 == i ? 0 : os[s])));
 160
 161                 //Update stacks
 162                 if (go[i] > 0)
 163                         Sdir.push(i);
 164                 else
 165                         Srev.push(i);
 166         }
 167         return components;
 168 }
 169
 170 /**
 171  *
 172  */
 173 std::vector<Interval> findIntervals(const GeneOrder& go){
 174         vector<Interval> intervals(go.size()-1,Interval(go.size(),go.size()));
 175         size_t n = 0;
 176         for (GeneOrder::iterator g = go.begin(); g != go.end(); ++g, ++n){
 177                         size_t i = abs(*g);
 178                         if (i < go.size() - 1)
 179                                 intervals[i].first = n + (*g >= 0 ? 1 : 0);
 180                         if (i > 0)
 181                                 intervals[i-1].second = n + (*g < 0 ? 1 : 0);
 182         }
 183         return intervals;
 184 }
 185
 186 /**
 187  *
 188  */
 189 std::vector<Interval> findIntervalsAtPoints(const vector<Interval>& intervals){
 190         size_t max = intervals.size()+1;
 191         vector<Interval> points(max,Interval(max,max));
 192         size_t n = 0;
 193         for (vector<Interval>::const_iterator i = intervals.begin(); i != intervals.end(); ++i, ++n){
 194                 if (points[i->first].first == max){
 195                         points[i->first].first = n;
 196                 }else
 197                         points[i->first].second = n;
 198
 199                 if (points[i->second].first == max){
 200                         points[i->second].first = n;
 201                 }else
 202                         points[i->second].second = n;
 203         }
 204         return points;
 205
 206 }