]> git.donarmstrong.com Git - mothur.git/blob - chimerarealigner.cpp
fixed bug with realigner for chimera slayer
[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 ms, int mm) : match(ms), misMatch(mm) {  templateSeqs = t;   m = MothurOut::getInstance(); }
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);  }
33                         int longest = 0;
34
35                         //take query and break apart into pieces using breakpoints given by results of parents
36                         for (int i = 0; i < parents.size(); i++) {
37                                 int length = parents[i].nastRegionEnd - parents[i].nastRegionStart+1;
38                                 string q = qAligned.substr(parents[i].nastRegionStart, length);
39         
40                                 Sequence* queryFrag = new Sequence(query->getName(), q);
41
42                                 queryParts.push_back(queryFrag);
43                         
44                                 Sequence* parent = getSequence(parents[i].parent);
45                                 string p = parent->getAligned();
46                 
47                                 p = p.substr(parents[i].nastRegionStart, length);
48                                 parent->setAligned(p);
49                                 
50                                 parentParts.push_back(parent);
51
52                                 if (queryFrag->getUnaligned().length() > longest)       { longest = queryFrag->getUnaligned().length(); }
53                                 if (parent->getUnaligned().length() > longest)  { longest = parent->getUnaligned().length();    }
54                         }
55
56                         //align each peice to correct parent from results
57                         for (int i = 0; i < queryParts.size(); i++) {
58                                 if ((queryParts[i]->getUnaligned() == "") || (parentParts[i]->getUnaligned() == "")) {;}
59                                 else {
60                                         Alignment* alignment = new NeedlemanOverlap(-2.0, 1.0, -1.0, longest+1); //default gapopen, match, mismatch, longestbase
61                                 
62                                         Nast nast(alignment, queryParts[i], parentParts[i]);
63                                         delete alignment;
64                                 }
65                         }
66
67                         //recombine pieces to form new query sequence
68                         for (int i = 0; i < queryParts.size(); i++) {
69                                 //sometimes the parent regions do not meet, for example region 1 may end at 1000 and region 2 starts at 1100.  
70                                 //we don't want to loose length so in this case we will leave query alone
71                                 if (i != 0) {
72                                         int space = parents[i].nastRegionStart - parents[i-1].nastRegionEnd - 1;
73                                         if (space > 0) { //they don't meet and we need to add query piece
74                                                 string q = qAligned.substr(parents[i-1].nastRegionEnd+1, space);
75                                                 newQuery += q;
76                                         }
77                                 }
78
79                                 newQuery += queryParts[i]->getAligned();
80                         }
81                         
82                         //make sure you don't cutoff end of query 
83                         if (parents[parents.size()-1].nastRegionEnd < (qAligned.length()-1)) {  newQuery += qAligned.substr(parents[parents.size()-1].nastRegionEnd+1);  }
84                         
85                         query->setAligned(newQuery);
86
87                         //free memory
88                         for (int i = 0; i < queryParts.size(); i++) { delete queryParts[i];  }
89                         for (int i = 0; i < parentParts.size(); i++) { delete parentParts[i];  }
90                         
91                 } //else leave query alone, you have bigger problems...
92                 
93         }
94         catch(exception& e) {
95                 m->errorOut(e, "ChimeraReAligner", "reAlign");
96                 exit(1);
97         }
98 }
99 //***************************************************************************************************************
100 Sequence* ChimeraReAligner::getSequence(string name) {
101         try{
102                 Sequence* temp;
103                 
104                 //look through templateSeqs til you find it
105                 int spot = -1;
106                 for (int i = 0; i < templateSeqs.size(); i++) {
107                         if (name == templateSeqs[i]->getName()) {  
108                                 spot = i;
109                                 break;
110                         }
111                 }
112                 
113                 if(spot == -1) { m->mothurOut("Error: Could not find sequence."); m->mothurOutEndLine(); return NULL; }
114                 
115                 temp = new Sequence(templateSeqs[spot]->getName(), templateSeqs[spot]->getAligned());
116                 
117                 return temp;
118         }
119         catch(exception& e) {
120                 m->errorOut(e, "ChimeraReAligner", "getSequence");
121                 exit(1);
122         }
123 }
124 //***************************************************************************************************************