]> git.donarmstrong.com Git - mothur.git/blob - chimerarealigner.cpp
working on chimeras
[mothur.git] / chimerarealigner.cpp
1 /*
2  *  chimerarealigner.cpp
3  *  Mothur
4  *
5  *  Created by westcott on 2/12/10.
6  *  Copyright 2010 Schloss Lab. All rights reserved.
7  *
8  */
9
10 #include "chimerarealigner.h"
11 #include "needlemanoverlap.hpp"
12 #include "nast.hpp"
13
14 //***************************************************************************************************************
15 ChimeraReAligner::ChimeraReAligner(vector<Sequence*> t, int m, int mm) : match(m), misMatch(mm) {  templateSeqs = t;  }
16 //***************************************************************************************************************
17 ChimeraReAligner::~ChimeraReAligner() {}        
18 //***************************************************************************************************************
19 void ChimeraReAligner::reAlign(Sequence* query, vector<results> parents) {
20         try {
21                 if (parents.size() != 0) {
22                         vector<Sequence*> queryParts;
23                         vector<Sequence*> parentParts;  //queryParts[0] relates to parentParts[0]
24                         
25                         string qAligned = query->getAligned();
26                         string newQuery = "";
27                         
28                         //sort parents by region start
29                         sort(parents.begin(), parents.end(), compareRegionStart);
30
31                         //make sure you don't cutoff beginning of query 
32                         if (parents[0].nastRegionStart > 0) {  newQuery += qAligned.substr(0, parents[0].nastRegionStart+1);  }
33                         int longest = 0;
34                         //take query and break apart into pieces using breakpoints given by results of parents
35                         for (int i = 0; i < parents.size(); i++) {
36                                 int length = parents[i].nastRegionEnd - parents[i].nastRegionStart+1;
37                                 string q = qAligned.substr(parents[i].nastRegionStart, length);
38                                 Sequence* queryFrag = new Sequence(query->getName(), q);
39                                 
40                                 queryParts.push_back(queryFrag);
41                                 
42                                 Sequence* parent = getSequence(parents[i].parent);
43                                 string p = parent->getAligned();
44                                 p = p.substr(parents[i].nastRegionStart, length);
45                                 parent->setAligned(p);
46                                 
47                                 parentParts.push_back(parent);
48                                 
49                                 if (q.length() > longest)       { longest = q.length(); }
50                                 if (p.length() > longest)       { longest = p.length(); }
51                         }
52                                                                                         
53                         //align each peice to correct parent from results
54                         for (int i = 0; i < queryParts.size(); i++) {
55                                 alignment = new NeedlemanOverlap(-2.0, match, misMatch, longest+1); //default gapopen, match, mismatch, longestbase
56                                 Nast nast(alignment, queryParts[i], parentParts[i]);
57                                 delete alignment;
58                         }
59                                                                                                 
60                         //recombine pieces to form new query sequence
61                         for (int i = 0; i < queryParts.size(); i++) {
62                                 newQuery += queryParts[i]->getAligned();
63                         }
64                         
65                         //make sure you don't cutoff end of query 
66                         if (parents[parents.size()-1].nastRegionEnd < qAligned.length()) {  newQuery += qAligned.substr(parents[parents.size()-1].nastRegionEnd-1);  }
67                         
68                         //set query to new aligned string
69                         query->setAligned(newQuery);
70                         
71                         //free memory
72                         for (int i = 0; i < queryParts.size(); i++) { delete queryParts[i];  }
73                         for (int i = 0; i < parentParts.size(); i++) { delete parentParts[i];  }
74                         
75                 } //else leave query alone, you have bigger problems...
76                 
77         }
78         catch(exception& e) {
79                 errorOut(e, "ChimeraReAligner", "reAlign");
80                 exit(1);
81         }
82 }
83 //***************************************************************************************************************
84 Sequence* ChimeraReAligner::getSequence(string name) {
85         try{
86                 Sequence* temp;
87                 
88                 //look through templateSeqs til you find it
89                 int spot = -1;
90                 for (int i = 0; i < templateSeqs.size(); i++) {
91                         if (name == templateSeqs[i]->getName()) {  
92                                 spot = i;
93                                 break;
94                         }
95                 }
96                 
97                 if(spot == -1) { mothurOut("Error: Could not find sequence."); mothurOutEndLine(); return NULL; }
98                 
99                 temp = new Sequence(templateSeqs[spot]->getName(), templateSeqs[spot]->getAligned());
100                 
101                 return temp;
102         }
103         catch(exception& e) {
104                 errorOut(e, "ChimeraReAligner", "getSequence");
105                 exit(1);
106         }
107 }
108 //***************************************************************************************************************