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