]> git.donarmstrong.com Git - mothur.git/blob - slayer.cpp
working on chimera.slayer
[mothur.git] / slayer.cpp
1 /*
2  *  slayer.cpp
3  *  Mothur
4  *
5  *  Created by westcott on 9/25/09.
6  *  Copyright 2009 Schloss Lab. All rights reserved.
7  *
8  */
9
10 #include "slayer.h"
11
12 /***********************************************************************/
13 Slayer::Slayer(int win, int increment, int parentThreshold, float div, int i, int snp, int mi) :
14                 minBS(mi), windowSize(win), windowStep(increment), parentFragmentThreshold(parentThreshold), divRThreshold(div), iters(i), percentSNPSample(snp){ m = MothurOut::getInstance(); }
15 /***********************************************************************/
16 string Slayer::getResults(Sequence* query, vector<Sequence*> refSeqs) {
17         try {
18                 vector<data_struct> all; all.clear();
19                 myQuery = *query;
20                 /*cout << "here" << endl;       
21                 query->setName("S000381746"); query->setAligned("...............................................................................................................................................A-C-GC--TGG-C--G-GC-A-GG--C----C-T--AACACA-T-GC-A-AGT-CGA-G-CG----------G-CAG-CG-G---------------------------GA-GG-A-AG----------------------------------------------------CTT-G----------------------------------------------------------------------------------CTT-CCTC----------------G-CC--G--GC--G--AG-C-GG-C-GG-A--C-------------GGG-TGAGT-A--AT-GT-C-T-G-GG---G-A--T-CT-G--C-C-CGA--TG-G------------------------------------------------------------------A-GG----GGG-AT-AA-CCA-------------------------C-T-G-----------------------GAA-A---CGG-TGG-CTAA-TA---CC-G--C-AT-A----------A--------------------C-------------------------------------GT-C-----------------------------------------------------------------------------------------------------------------------G-CA-A--------------------------------------------------------------------------------------------------------------------------------------G-A-C---------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------CAAA--G-T-G-GG-----G--GA-C--C--------------------------------------------------------------------------------------------------------------------TTC-G----------------------------------------------------------------------------------------------------------------------G-G--CC-TC--A---C-A--------------C----C-A---T-CG-G---AT---G-A-----A-CCC-AGA--T-GGG--A------TT--A--G-CT-A----G---TAGG-T-G-GG-G-T----AAT-GG-C-T-C-ACCT--A-GG-C-G--A-CG-A------------TCC-C-T------AG-CT-G-G-TCT-G-AG----A--GG-AT--G-AC-C-AG-CCAC-A-CTGGA--A-C-TG-A-GA-C-AC-G-G-TCCAGA-CTCC-TAC-G--G-G-A-G-GC-A-GC-A-G-TG---GG-G-A-ATA-TTGCA-C-AA-T-GG--GC-GC-A----A-G-CC-T-GA-TG-CA-GCCA-TGCC-G-CG-T---G-T-G--T--GA-A-G--A--A-G-G-CC-----TT-CG---------G-G-T-T-G-T--A---AA-G-CAC--------TT-TC-A-G--C-GAG----GA-G--G---AA-GGTG---GTGA-GC----T--T--AA-T---A----------CG-CTCAT-CAA-TT-GA-CG-TT-A-C-TC-G-CA-G---------AA-----------GAAGC-ACC-GG-C-TAA---C--T-CCGT--GCCA--G-C---A--GCCG---C-GG--TA-AT--AC---GG-AG-GGT-GCA-A-G-CG-TTAA-T-CGG-AA-TT-A--C-T--GGGC-GTA----AA-GCGC-AC--G-CA-G-G-C-G------------G--T-TT-G-T-T-AA----G-T-C-A---G-ATG-TG-A-AA-TC--CC-CGA-G--------------------------------------------------------------------CT-T-AA-------------------------------------------------------------------------CT-T-G-GG-AA-C----T-G-C-A-T-T--------T--GA-A-A-C-T-G-GCA--A-G-C---------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------T-A-G-A-G-T-C-----T-CG--TA-G-A------------G-GG-G-GG-T----AG--AATT-CCA-G-GT--GT-A-GCG-GTGAAA-TG-CGT-AGAG-A-TC-T-GGA--GG-A-AT-A-CC-GG--T--G--GC-GAA-G--G-C---G----G--C-C-CCCTG------G-AC-GA--------------------------------------------------------------AG-A-C-T--GA--CG-----CT-CA-GG--T-G-CGA--AA-G-C--------------G-TGGG-GAG-C-A-AACA--GG-ATTA-G-ATA-C-----CC-T-G-GTA-G-T----C-CA--C-G-CCG-T-AAA--C-GATG-TC--GA-TT---------T-GG--A--G-G-TT-G-TG-C--C--------------------------------------------------------------------------------------CTT-GA--------------------------------------------------------------------------------------------------------------------------------------------------G-G-C-GT--G-G-C-T-TC-C------GG--A----GC-TAA--CG-C-G-T--T--AA-AT--C----G-ACC-GCC-T-G-GG-GAG-TA---CGG-----C-C--G-C-A-A-GGT-T--AAA-ACTC-AAA---------TGAA-TTG-ACGGG-G-G-CCCG----C-A--C-A-A-GCG-GT-G--G--AG-CA-T--GT-GGT-TT-AATT-C-G-ATG-CAAC-G-CG-A-AG-A-A-CC-TT-A-CC-TACTC-TT-G-AC-A-T-C--------------CAG-A-G-------------A-AC-T-T-T--CC--A-GA-G-A-T--G-G-A--T-T-G-G--T-G-----CC-------------------------------------T--TC-G------------------------------------------GG----A----A---CT-CTG---A--GA---------------------------------------------------C-A-G-G-T-GCTG-CA-TGG-CT--GTC-GTC-A-GC-TC---G-TG-TT-G--TGA-AA-TGT-T-GG-G-TT-AA-GT-CCCGC-AA--------C-GAG-CGC-A-ACC-C-T-TA--TC--C-TTTG--T-T-G-C-C---AG-C-G--G--T-------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------TCG------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------G----C----C-G------------G----G---A-A--CT---------------C-A-A-A-G-GA-G--AC-T-G-CCA--G-T------------------------------------G-A---TAA----------------------------------A-C-T-G--G-A-GG-A--AGG-T--GGGG-A-TGAC-GTC--AAGT-C---ATC-A-T-G-G-C-C-CTT----AC-G--AG-T-A-GG-GC-TA-CAC-ACGTG-C--TA--CAATG---G-CGTA-T-A--C-AAA-GA-GA--------------------------------------------------------------------------------------------------A-G-C-G-A--ACCT-G-C--G---------------------------------------A-GG-G-C-----------A--A-G-CG---G----------A--CCT-C------A-T-AAAGT-AC-G-T-C-G-TAG-TCC--------GGA-T-TGGAG-TC--T-GCAA-CT-C-------------------------------------------------------------------------------------------------G-ACTCC-A-T-G-AA-G-TC-GGAAT-CG-C-TA--G-TA-AT-C-G-T----AGA-TC-A-G--A------AT--GCT-AC-G-GT-G-AAT-ACGT-T-CCCGGGCCT-TGTA----CACACCG-CCC-GTC-----A---CA--CCA-TG-GG-A--G---TGG-G-TT-GC-AAA--A-GAA------G--T-AGG-TA-G-C-T-T-AA-C-C-------------------------------------------------------------T-TC-G------------------------------------------------------------------------------------------------------GG-A--GG-G--C---GC-TTA--CC--ACT-T----T-GT..........................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................");
22                 refSeqs.resize(2);
23                 refSeqs[1]->setName("S000381740"); refSeqs[1]->setAligned("...............................................................................................................................................a-c-gc--tgg-c--g-gc-a-gg--c----c-t--aacaca-t-gc-a-agt-cga-g-cg----------g-tag-ca-c----------------------------agga-g-ag----------------------------------------------------ctt-g----------------------------------------------------------------------------------ctc-tctg----------------g-gt--g--ac--g--ag-c-gg-c-gg-a--c-------------ggg-tgagt-a--at-gt-c-t-g-gg---a-a--a-ct-g--c-c-tga--tg-g------------------------------------------------------------------a-gg----ggg-at-aa-cta-------------------------c-t-g-----------------------gaa-a---cgg-tag-ctaa-ta---cc-g--c-at-a----------a--------------------c-------------------------------------gt-c-----------------------------------------------------------------------------------------------------------------------t-ac-g--------------------------------------------------------------------------------------------------------------------------------------g-a-c---------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------caaa--g-t-g-gg-----g--ga-c--c--------------------------------------------------------------------------------------------------------------------ttc-g----------------------------------------------------------------------------------------------------------------------g-g--cc-tc--a---c-g--------------c----c-a---t-ca-g---at---g-t-----g-ccc-aga--t-ggg--a------tt--a--g-ct-a----g---tagg-t-g-gg-g-t----aat-gg-c-t-c-acct--a-gg-c-g--a-cg-a------------tcc-c-t------ag-ct-g-g-tct-g-ag----a--gg-at--g-ac-c-ag-ccac-a-ctgga--a-c-tg-a-ga-c-ac-g-g-tccaga-ctcc-tac-g--g-g-a-g-gc-a-gc-a-g-tg---gg-g-a-ata-ttgca-c-aa-t-gg--gc-gc-a----a-g-cc-t-ga-tg-ca-gcca-tgcc-g-cg-t---g-t-g--t--ga-a-g--a--a-g-g-cc-----tt-cg---------g-g-t-t-g-t--a---aa-g-cac--------tt-tc-a-g--c-gag----ga-g--g---aa-gggc---gatg-tc----t--t--aa-t---a----c-----gg-c-agc-gca-tt-ga-cg-tt-a-c-tc-g-ca-g---------aa-----------gaagc-acc-gg-c-taa---c--t-ccgt--gcca--g-c---a--gccg---c-gg--ta-at--ac---gg-ag-ggt-gca-a-g-cg-ttaa-t-cgg-aa-tt-a--c-t--gggc-gta----aa-gcgc-ac--g-ca-g-g-c-g------------g--t-tt-g-t-t-aa----g-t-c-a---g-atg-tg-a-aa-tc--cc-cgc-g--------------------------------------------------------------------ct-t-aa-------------------------------------------------------------------------cg-t-g-gg-aa-c----t-g-c-a-t-t--------t--ga-a-a-c-t-g-gca--a-g-c---------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------t-a-g-a-g-t-c-----t-cg--ta-g-a------------g-gg-g-gg-t----ag--aatt-cca-g-gt--gt-a-gcg-gtgaaa-tg-cgt-agag-a-tc-t-gga--gg-a-at-a-cc-gg--t--g--gc-gaa-g--g-c---g----g--c-c-ccctg------g-ac-ga--------------------------------------------------------------ag-a-c-t--ga--cg-----ct-ca-gg--t-g-cga--aa-g-c--------------g-tggg-gag-c-a-aaca--gg-atta-g-ata-c-----cc-t-g-gta-g-t----c-ca--c-g-ctg-t-aaa--c-gatg-tc--ga-tt---------t-gg--a--g-g-tt-g-tg-c--c--------------------------------------------------------------------------------------ctt-ga--------------------------------------------------------------------------------------------------------------------------------------------------g-g-c-gt--g-g-c-t-tc-c------gg--a----gc-taa--cg-c-g-t--t--aa-at--c----g-acc-gcc-t-g-gg-gag-ta---cgg-----c-c--g-c-a-a-ggt-t--aaa-actc-aaa---------tgaa-ttg-acggg-g-g-cccg----c-a--c-a-a-gcg-gt-g--g--ag-ca-t--gt-ggt-tt-aatt-c-g-atg-caac-g-cg-a-ag-a-a-cc-tt-a-cc-tactc-tt-g-ac-a-t-c--------------cag-a-g-------------a-ac-t-t-t--cc--a-ga-g-a-t--g-g-a--t-t-g-g--t-g-----cc-------------------------------------t--tc-g------------------------------------------gg----a----a---ct-ctg---a--ga---------------------------------------------------c-a-g-g-t-gctg-ca-tgg-ct--gtc-gtc-a-gc-tc---g-tg-tt-g--tga-aa-tgt-t-gg-g-tt-aa-gt-cccgc-aa--------c-gag-cgc-a-acc-c-t-ta--tc--c-tttg--t-t-g-c-c---ag-c-g--a--t-------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------tcg------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------g----t----c-g------------g----g---a-a--ct---------------c-a-a-a-g-ga-g--ac-t-g-ccg--g-t------------------------------------g-a---taa----------------------------------a-c-c-g--g-a-gg-a--agg-t--gggg-a-tgac-gtc--aagt-c---atc-a-t-g-g-c-c-ctt----ac-g--ag-t-a-gg-gc-ta-cac-acgtg-c--ta--caatg---g-cgta-t-a--c-aaa-ga-ga--------------------------------------------------------------------------------------------------a-g-c-g-a--a-ctcg-c--g---------------------------------------a-ga-g-c-----------a--a-g-cg---g----------a--cct-c------a-t-aaagt-ac-g-t-c-g-tag-tcc--------gga-t-tggag-tc--t-gcaa-ct-c-------------------------------------------------------------------------------------------------g-actcc-a-t-g-aa-g-tc-ggaat-cg-c-ta--g-ta-at-c-g-t----aga-tc-a-g--a------at--gct-ac-g-gt-g-aat-acgt-t-cccgggcct-tgta----cacaccg-ccc-gtc-----a---ca--cca-tg-gg-a--g---tgg-g-tt-gc-aaa--a-gaa------g--t-agg-ta-g-c-t-t-aa-c-c-------------------------------------------------------------t-tc-g------------------------------------------------------------------------------------------------------gg-a--gg-g--c---gc-tta--cc--act-t----t-gtg-at-tca------------------------t.........................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................");
24                 refSeqs[0]->setName("7000004131500404"); refSeqs[0]->setAligned(".........................................................................................................................................AT-TGAA-C-GC--TGG-C--G-GC-A-GG--C----C-T--AACACA-T-GC-A-AGT-CGA-G-CG----------G-CAG-CG-G----------------------------AAAG-A-AG----------------------------------------------------CTT-G---------------------------------------------------------------------------------ACTT-CTTT----------------G-CC--G--GC--G--AG-C-GG-C-GG-A--C-------------GGG-TGAGT-A--AT-GT-C-T-G-GG---G-A--T-CT-G--C-C-CGA--TG-G------------------------------------------------------------------A-GG----GGG-AT-AA-CTA-------------------------C-T-G-----------------------GAA-A---CGG-TAG-CTAA-TA---CC-G--C-AT-A----------A--------------------C-------------------------------------GT-C-----------------------------------------------------------------------------------------------------------------------G-CA-A--------------------------------------------------------------------------------------------------------------------------------------G-A-C---------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------CAAA--G-T-G-GG-----G--GA-C--C--------------------------------------------------------------------------------------------------------------------TTC-G----------------------------------------------------------------------------------------------------------------------G-G--CC-TC--A---C-A--------------C----C-A---T-CG-G---AT---G-A-----A-CCC-AGA--T-GGG--A------TT--A--G-CT-A----G---TAGG-T-G-GG-G-T----AAT-GG-C-T-C-ACCT--A-GG-C-G--A-CG-A------------TCC-C-T------AG-CT-G-G-TCT-G-AG----A--GG-AT--G-AC-C-AG-CCAC-A-CTGGA--A-C-TG-A-GA-C-AC-G-G-TCCAGA-CTCC-TAC-G--G-G-A-G-GC-A-GC-A-G-TG---GG-G-A-ATA-TTGCA-C-AA-T-GG--GG-GA-A----A-C-CC-T-GA-TG-CA-GCCA-TGCC-G-CG-T---G-T-G--T--GA-A-G--A--A-G-G-CC-----TT-CG---------G-G-T-T-G-T--A---AA-G-CAC--------TT-TC-A-G--C-GGG----GA-A--G---AA-GGCG---TT-A-GC---GT--T--AA-C---A----G-----CG-C-TAT-CGA-TT-GA-CG-TT-A-C-CT-G-CA-G---------AA-----------GAAGC-ACC-GG-C-TAA---C--T-CCGT--GCCA--G-C---A--GCCG---C-GG--TA-AT--AC---GG-AG-GGT-GCA-A-G-CG-TTAA-T-CGG-AA-TT-A--C-T--GGGC-GTA----AA-GCGT-AC--G-CA-G-G-C-G------------G--T-CT-G-T-T-AA----G-T-C-A---G-ATG-TG-A-AA-TC--CC-CGG-G--------------------------------------------------------------------CT-T-AA-------------------------------------------------------------------------CC-T-G-GG-AA-C----T-G-C-A-T-T--------T--GA-A-A-C-T-G-GCA--G-G-C---------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------T-A-G-A-G-T-C-----T-CG--TA-G-A------------G-GG-G-GG-T----AG--AATT-CCA-G-GT--GT-A-GCG-GTGAAA-TG-CGT-AGAG-A-TC-T-GGA--GG-A-AT-A-CC-GG--T--G--GC-GAA-G--G-C---G----G--C-C-CCCTG------G-AC-GA--------------------------------------------------------------AG-A-C-T--GA--CG-----CT-CA-GG--T-A-CGA--AA-G-C--------------G-TGGG-GAG-C-A-AACA--GG-ATTA-G-ATA-C-----CC-T-G-GTA-G-T----C-CA--C-G-CTG-T-AAA--C-GATG-TC--GA-TT---------T-GA--A--G-G-TT-G-TG-G--C--------------------------------------------------------------------------------------CTT-GA--------------------------------------------------------------------------------------------------------------------------------------------------G-C-T-GT--G-G-C-T-TT-C------GG--A----GC-TAA--CG-C-G-T--T--AA-AT--C----G-ACC-GCC-T-G-GG-GAG-TA---CGG-----C-C--G-C-A-A-GGT-T--AAA-ACTC-AAA---------TGAA-TTG-ACGGG-G-G-CCCG----C-A--C-A-A-GCG-GT-G--G--AG-CA-T--GT-GGT-TT-AATT-C-G-ATG-CAAC-G-CG-A-AG-A-A-CC-TT-A-CC-TACTC-TT-G-AC-A-T-C--------------CAG-A-G-------------A-AC-T-T-G--GC--A-GA-G-A-T--G-C-C--T-T-G-G--T-G-----CC-------------------------------------T--TC-G------------------------------------------GG----A----G---CT-CTG---A--GA---------------------------------------------------C-A-G-G-T-GCTG-CA-TGG-CT--GTC-GTC-A-GC-TC---G-TG-TT-G--TGA-AA-TGT-T-GG-G-TT-AA-GT-CCCGC-AA--------C-GAG-CGC-A-ACC-C-T-TA--TC--C-TTTG--T-T-G-C-C---AG-C-G--A--T-------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------TTG------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------G----T----C-G------------G----G---A-A--CT---------------C-A-A-A-G-GA-G--AC-T-G-CCG--G-T------------------------------------G-A---TAA----------------------------------A-C-C-G--G-A-GG-A--AGG-T--GGGG-A-TGAC-GTC--AAGT-C---ATC-A-T-G-G-C-C-CTT----AC-G--AG-T-A-GG-GC-TA-CAC-ACGTG-C--TA--CAATG---G-CGCA-T-A--C-AAA-GA-GA--------------------------------------------------------------------------------------------------A-G-C-G-A--T-CTCG-C--G---------------------------------------A-GA-G-T-----------C--A-G-CG---G----------A--CCT-C------A-C-AAAGT-GC-G-T-C-G-TAG-TCC--------GGA-T-TGGAG-TC--T-GCAA-CT-C-------------------------------------------------------------------------------------------------G-ACTCC-A-T-G-AA-G-TC-GGAAT-CG-C-TA--G-TA-AT-C-G-T----GGA-TC-A-G--A------AT--GCC-AC-G-GT-G-AAT-ACGT-T-CCTGGGCCT-TGTA----CACACCG-CCC-GTC-----A---CA--CCA-TG-GG-A--G---TGG-G-TT-GC-AAA--A-GAA------G--T-AGG-TA-G-C-T-T-AA-C-C-------------------------------------------------------------T-TC-G------------------------------------------------------------------------------------------------------GG-A--GG-G--C---GC-TTA--CC--ACT-T----T-GTG-AT-TCA------------------------TG--ACT-GGGG-TG-AAG-TCGTAACAA-GGTAA-CCGT-AGGGGAA-CCT......................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................");          
25                 for (int i = 0; i < refSeqs.size(); i++) {
26                         string newAligned = "";
27                         for (int j = 0; j < refSeqs[i]->getAligned().length(); j++) {
28                                 newAligned += toupper(refSeqs[i]->getAligned()[j]);
29                         }
30                         refSeqs[i]->setAligned(newAligned);
31                 }*/
32                 
33                 for (int i = 0; i < refSeqs.size(); i++) {
34                 
35                         for (int j = i+1; j < refSeqs.size(); j++) {
36                         
37                                 if (m->control_pressed) { return "no";  }
38         
39                                 //make copies of query and each parent because runBellerophon removes gaps and messes them up
40                                 Sequence* q = new Sequence(query->getName(), query->getAligned());
41                                 Sequence* leftParent = new Sequence(refSeqs[i]->getName(), refSeqs[i]->getAligned());
42                                 Sequence* rightParent = new Sequence(refSeqs[j]->getName(), refSeqs[j]->getAligned());
43
44                                 map<int, int> spots;  //map from spot in original sequence to spot in filtered sequence for query and both parents
45                                 vector<data_struct> divs = runBellerophon(q, leftParent, rightParent, spots);
46                                 //cout << divs.size() << endl;
47                                 if (m->control_pressed) { 
48                                         delete q;
49                                         delete leftParent;
50                                         delete rightParent;
51                                         return "no"; 
52                                 }
53                                         
54 //                              cout << divs.size() << endl;
55                                 vector<data_struct> selectedDivs;
56                                 for (int k = 0; k < divs.size(); k++) {
57                                         
58                                         vector<snps> snpsLeft = getSNPS(divs[k].parentA.getAligned(), divs[k].querySeq.getAligned(), divs[k].parentB.getAligned(), divs[k].winLStart, divs[k].winLEnd);
59                                         vector<snps> snpsRight = getSNPS(divs[k].parentA.getAligned(), divs[k].querySeq.getAligned(), divs[k].parentB.getAligned(), divs[k].winRStart, divs[k].winREnd);
60                                         
61                                         if (m->control_pressed) { 
62                                                 delete q;
63                                                 delete leftParent;
64                                                 delete rightParent;
65                                                 return "no"; 
66                                         }
67                                         
68                                         int numSNPSLeft = snpsLeft.size();
69                                         int numSNPSRight = snpsRight.size();
70                                         
71                                         //require at least 3 SNPs on each side of the break
72 //                                      if ((numSNPSLeft >= 3) && (numSNPSRight >= 3)) {
73                                         if ((numSNPSLeft >= 4) && (numSNPSRight >= 4)) {
74                                         
75                                                 //removed in 12/09 version of chimeraSlayer
76                                                 //int winSizeLeft = divs[k].winLEnd - divs[k].winLStart + 1;
77                                                 //int winSizeRight = divs[k].winREnd - divs[k].winRStart + 1;
78                                                 
79                                                 //float snpRateLeft = numSNPSLeft / (float) winSizeLeft;
80                                                 //float snpRateRight = numSNPSRight / (float) winSizeRight;
81                                                 //float logR = log(snpRateLeft / snpRateRight) / log(2.0); 
82                                                 
83                                                 // do not accept excess snp ratio on either side of the break
84                                                 //if (abs(logR) < 1 ) {  
85                                                         
86                                                         float BS_A, BS_B;
87                                                         bootstrapSNPS(snpsLeft, snpsRight, BS_A, BS_B, iters);
88                                                         
89                                                         if (m->control_pressed) { 
90                                                                 delete q;
91                                                                 delete leftParent;
92                                                                 delete rightParent;
93                                                                 return "no"; 
94                                                         }
95
96                                                         divs[k].bsa = BS_A;
97                                                         divs[k].bsb = BS_B;
98                                                 
99                                                         divs[k].bsMax = max(BS_A, BS_B);
100                                                 
101                                                         divs[k].chimeraMax = max(divs[k].qla_qrb, divs[k].qlb_qra);
102                                                         
103                                                         //so results reflect orignal alignment
104                                                         divs[k].winLStart = spots[divs[k].winLStart];
105                                                         divs[k].winLEnd = spots[divs[k].winLEnd];  
106                                                         divs[k].winRStart = spots[divs[k].winRStart]; 
107                                                         divs[k].winREnd = spots[divs[k].winREnd]; 
108                                                 
109                                                         selectedDivs.push_back(divs[k]);
110                                                 //}
111                                         }
112                                 }
113                                 
114                                 //save selected
115                                 for (int m = 0; m < selectedDivs.size(); m++) {  all.push_back(selectedDivs[m]);        }
116                                 
117                                 delete q;
118                                 delete leftParent;
119                                 delete rightParent;
120                         }
121                 }
122                 
123                 // compute bootstrap support
124                 if (all.size() > 0) {
125                         //sort them
126                         sort(all.begin(), all.end(), compareDataStruct);
127                         reverse(all.begin(), all.end());
128                         
129                         outputResults = all;
130                         return "yes"; 
131                 }
132                 else {
133                         outputResults = all;
134                         return "no";
135                 }
136         }
137         catch(exception& e) {
138                 m->errorOut(e, "Slayer", "getResults");
139                 exit(1);
140         }
141 }
142 /***********************************************************************/
143 vector<data_struct> Slayer::runBellerophon(Sequence* q, Sequence* pA, Sequence* pB, map<int, int>& spots) {
144         try{
145                 
146                 vector<data_struct> data;
147                 
148                 //vertical filter
149                 vector<Sequence*> temp;
150                 temp.push_back(q); temp.push_back(pA); temp.push_back(pB);
151                 
152                 //maps spot in new alignment to spot in alignment before filter
153                 spots = verticalFilter(temp);  //fills baseSpots
154                 
155                 //get these to avoid numerous function calls
156                 string query = q->getAligned();
157                 string parentA = pA->getAligned();
158                 string parentB = pB->getAligned();
159                 int length = query.length();
160 /*cout << q->getName() << endl << q->getAligned() << endl << endl;      
161 cout << pA->getName() << endl << pA->getAligned() << endl << endl;              
162 cout << pB->getName() << endl << pB->getAligned() << endl << endl;      
163 cout << " length = " << length << endl;
164 cout << q->getName() << endl;
165 cout << pA->getName() << '\t';
166 cout << pB->getName() << endl;*/
167         
168                 //check window size
169                 if (length < (2*windowSize+windowStep)) { 
170 //                      m->mothurOut("Your window size is too large for " + q->getName() + ". I will make the window size " + toString(length/4) + " which is 1/4 the filtered length."); m->mothurOutEndLine();        
171                         windowSize = length / 4;
172                 }
173                 
174                 for (int i = windowSize-1; i <= (length - windowSize); i += windowStep) {
175                 
176                         if (m->control_pressed) { return data; }
177                 
178                         int breakpoint = i;
179                         int leftLength = breakpoint + 1;
180                         int rightLength = length - leftLength;
181                         
182                         float QLA = computePercentID(query, parentA, 0, breakpoint);
183                         float QRB = computePercentID(query, parentB, breakpoint+1, length - 1);
184                 
185                         float QLB = computePercentID(query, parentB, 0, breakpoint);
186                         float QRA = computePercentID(query, parentA, breakpoint+1, length - 1);
187                 
188                         float LAB = computePercentID(parentA, parentB, 0, breakpoint);
189                         float RAB = computePercentID(parentA, parentB, breakpoint+1, length - 1);
190                 
191                         float AB = ((LAB*leftLength) + (RAB*rightLength)) / (float) length;
192                         float QA = ((QLA*leftLength) + (QRA*rightLength)) / (float) length;
193                         float QB = ((QLB*leftLength) + (QRB*rightLength)) / (float) length;
194                 
195                         float QLA_QRB = ((QLA*leftLength) + (QRB*rightLength)) / (float) length;
196                         float QLB_QRA = ((QLB*leftLength) + (QRA*rightLength)) / (float) length;
197                 
198                         //in original and not used
199                         //float avgQA_QB = ((QA*leftLength) + (QB*rightLength)) / (float) length;
200                 
201                         float divR_QLA_QRB = min((QLA_QRB/QA), (QLA_QRB/QB));
202                         float divR_QLB_QRA = min((QLB_QRA/QA), (QLB_QRA/QB));
203 //cout << leftLength << '\t' << rightLength << '\t' << QLA << '\t' << QRB << '\t' << QLB << '\t' << QRA  << '\t' << LAB << '\t' << RAB << '\t' << AB << '\t' << QA << '\t' << QB << '\t' << QLA_QRB << '\t' <<  QLB_QRA <<    endl;             
204
205 //cout << divRThreshold << endl;
206 //cout << breakpoint << '\t' << divR_QLA_QRB << '\t' << divR_QLB_QRA << endl;
207                         //is one of them above the 
208                         if (divR_QLA_QRB >= divRThreshold || divR_QLB_QRA >= divRThreshold) {
209                                 
210                                 if (((QLA_QRB > QA) && (QLA_QRB > QB) && (QLA >= parentFragmentThreshold) && (QRB >= parentFragmentThreshold))  ||
211                                         ((QLB_QRA > QA) && (QLB_QRA > QB) && (QLB >=parentFragmentThreshold) && (QRA >= parentFragmentThreshold)))  {
212                                         
213                                         data_struct member;
214                                         
215                                         member.divr_qla_qrb = divR_QLA_QRB;
216                                         member.divr_qlb_qra = divR_QLB_QRA;
217                                         member.qla_qrb = QLA_QRB;
218                                         member.qlb_qra = QLB_QRA;
219                                         member.qla = QLA;
220                                         member.qrb = QRB;
221                                         member.ab = AB; 
222                                         member.qa = QA;
223                                         member.qb = QB; 
224                                         member.lab = LAB; 
225                                         member.rab = RAB; 
226                                         member.qra = QRA; 
227                                         member.qlb = QLB; 
228                                         member.winLStart = 0;
229                                         member.winLEnd = breakpoint;  
230                                         member.winRStart = breakpoint+1; 
231                                         member.winREnd = length-1; 
232                                         member.querySeq = *(q); 
233                                         member.parentA = *(pA);
234                                         member.parentB = *(pB);
235                                         member.bsa = 0;
236                                         member.bsb = 0;
237                                         member.bsMax = 0;
238                                         member.chimeraMax = 0;
239                                         
240                                         data.push_back(member);
241                                         
242                                 }//if
243                         }//if
244                 }//for
245                 
246                 return data;
247                 
248         }
249         catch(exception& e) {
250                 m->errorOut(e, "Slayer", "runBellerophon");
251                 exit(1);
252         }
253 }
254 /***********************************************************************/
255 vector<snps> Slayer::getSNPS(string parentA, string query, string parentB, int left, int right) {
256         try {
257         
258                 vector<snps> data;
259 //cout << left << '\t' << right << endl;
260                 for (int i = left; i <= right; i++) {
261                         
262                         char A = parentA[i];
263                         char Q = query[i];
264                         char B = parentB[i];
265                         
266                         if ((A != Q) || (B != Q)) {
267 //cout << "not equal " << Q << '\t' << A << '\t' << B << endl;
268                         
269                                 //ensure not neighboring a gap. change to 12/09 release of chimeraSlayer - not sure what this adds, but it eliminates alot of SNPS
270                                 if (
271                                         //did query loose a base here during filter??
272                                         ( i == 0 || abs (baseSpots[0][i] - baseSpots[0][i-1]) == 1) &&
273                                         ( i == query.length() || abs (baseSpots[0][i] - baseSpots[0][i+1]) == 1)
274                                         &&
275                                         //did parentA loose a base here during filter??
276                                         ( i == 0 || abs (baseSpots[1][i] - baseSpots[1][i-1]) == 1) &&
277                                         ( i == parentA.length() || abs (baseSpots[1][i] - baseSpots[1][i+1]) == 1) 
278                                         &&
279                                         //did parentB loose a base here during filter??
280                                         ( i == 0 || abs (baseSpots[2][i] - baseSpots[2][i-1]) == 1) &&
281                                         ( i == parentB.length() || abs (baseSpots[2][i] - baseSpots[2][i+1]) == 1)
282                                         ) 
283                                 { 
284                                 
285                                         snps member;
286                                         member.queryChar = Q;
287                                         member.parentAChar = A;
288                                         member.parentBChar = B;
289 //cout << "not neighboring a gap " << Q << '\t' << A << '\t' << B << '\t' << baseSpots[0][i] << '\t' << baseSpots[0][i+1] << '\t' << baseSpots[0][i-1] << '\t' << baseSpots[1][i] << '\t' << baseSpots[1][i+1] << '\t' << baseSpots[1][i-1] << '\t' << baseSpots[2][i] << '\t' << baseSpots[2][i+1] << '\t' << baseSpots[2][i-1] << endl;                               
290                                         data.push_back(member);
291                                 }
292                         }
293 //                      cout << i << '\t' << data.size() << endl;
294                 }
295                 
296                 return data;
297                 
298         }
299         catch(exception& e) {
300                 m->errorOut(e, "Slayer", "getSNPS");
301                 exit(1);
302         }
303 }
304 /***********************************************************************/
305 int Slayer::bootstrapSNPS(vector<snps> left, vector<snps> right, float& BSA, float& BSB, int numIters) {
306         try {
307
308                 srand((unsigned)time( NULL ));
309
310                 int count_A = 0; // sceneario QLA,QRB supported
311                 int count_B = 0; // sceneario QLB,QRA supported
312         
313                 int numLeft = max(1, int(left.size() * percentSNPSample/(float)100 + 0.5));
314                 int numRight = max(1, int(right.size() * percentSNPSample/(float)100 + 0.5));
315
316                 for (int i = 0; i < numIters; i++) {
317                         //random sampling with replacement.
318                 
319                         if (m->control_pressed) { return 0;  }
320                         
321                         vector<snps> selectedLeft;
322
323                         for (int j = 0; j < numLeft; j++) {
324                                 int index = int(rand() % left.size());
325                                 selectedLeft.push_back(left[index]);
326                         }
327
328                         vector<snps> selectedRight;
329                         for (int j = 0; j < numRight; j++) {
330                                 int index = int(rand() % right.size());
331                                 selectedRight.push_back(right[index]);
332                         }
333                 
334                         /* A  ------------------------------------------
335                         #       QLA                     QRA
336                         # Q  ------------------------------------------
337                         #                      |
338                         #                      |
339                         # Q  ------------------------------------------
340                         #       QLB                     QRB
341                         # B  ------------------------------------------ */
342                 
343                 
344                         float QLA = snpQA(selectedLeft);
345                         float QRA = snpQA(selectedRight);
346                 
347                         float QLB = snpQB(selectedLeft);
348                         float QRB = snpQB(selectedRight);
349         
350                         //in original - not used - not sure why?
351                         //float ALB = snpAB(selectedLeft);
352                         //float ARB = snpAB(selectedRight);
353                 
354                         if ((QLA > QLB) && (QRB > QRA)) {
355                                 count_A++;
356                         }
357                 
358                         if ((QLB > QLA) && (QRA > QRB)) {
359                                 count_B++;
360                         }
361                         
362 //cout << "selected left snp: \n";
363 //for (int j = 0; j < selectedLeft.size(); j++) {  cout << selectedLeft[j].parentAChar;  } 
364 //cout << endl;
365 //for (int j = 0; j < selectedLeft.size(); j++) {  cout << selectedLeft[j].queryChar;  }
366 //cout << endl;
367 //for (int j = 0; j < selectedLeft.size(); j++) {  cout << selectedLeft[j].parentBChar;  }
368 //cout << endl;
369 //cout << "selected right snp: \n";
370 //for (int j = 0; j < selectedRight.size(); j++) {  cout << selectedRight[j].parentAChar;  } 
371 //cout << endl;
372 //for (int i = 0; i < selectedRight.size(); i++) {  cout << selectedRight[i].queryChar;  }
373 //cout << endl;
374 //for (int i = 0; i < selectedRight.size(); i++) {  cout << selectedRight[i].parentBChar;  }
375 //cout << endl;         
376                 }
377
378
379
380
381                 BSA = (float) count_A / (float) iters * 100;
382                 BSB = (float) count_B / (float) iters * 100;
383 //cout << "bsa = " << BSA << " bsb = " << BSB << endl;
384                 
385                 //run borderline bootstrap values longer
386                 //if (numIters < 1000) {
387                         //are you within 10 points of min bootstrap value cutoff
388                 //      if (((abs((double)(BSA - minBS))) <= 5) || ((abs((double)(BSB - minBS))) <= 5)) {
389                 //              m->mothurOut("extending bootstrap for " + myQuery.getName()); m->mothurOutEndLine();
390                 //              bootstrapSNPS(left, right, BSA, BSB, 1000);
391                 //      }
392                 //}
393
394                 return 0;
395         
396         }
397         catch(exception& e) {
398                 m->errorOut(e, "Slayer", "bootstrapSNPS");
399                 exit(1);
400         }
401 }
402 /***********************************************************************/
403 float Slayer::snpQA(vector<snps> data) {
404         try {
405         
406                 int numIdentical = 0;
407         
408                 for (int i = 0; i < data.size(); i++) {
409                         if (data[i].parentAChar == data[i].queryChar) {
410                                 numIdentical++;
411                         }
412                 }
413
414                 float percentID = (numIdentical / (float) data.size()) * 100;
415                 
416                 return percentID;
417         }
418         catch(exception& e) {
419                 m->errorOut(e, "Slayer", "snpQA");
420                 exit(1);
421         }
422 }
423 /***********************************************************************/
424 float Slayer::snpQB(vector<snps> data) {
425         try {
426         
427                 int numIdentical = 0;
428         
429                 for (int i = 0; i < data.size(); i++) {
430                         if (data[i].parentBChar == data[i].queryChar) {
431                                 numIdentical++;
432                         }
433                 }
434
435                 float percentID = (numIdentical / (float) data.size()) * 100;
436                 
437                 return percentID;
438
439         }
440         catch(exception& e) {
441                 m->errorOut(e, "Slayer", "snpQB");
442                 exit(1);
443         }
444 }
445 /***********************************************************************/
446 float Slayer::snpAB(vector<snps> data) {
447         try {
448                 int numIdentical = 0;
449         
450                 for (int i = 0; i < data.size(); i++) {
451                         if (data[i].parentAChar == data[i].parentBChar) {
452                                 numIdentical++;
453                         }
454                 }
455
456                 float percentID = (numIdentical / (float) data.size()) * 100;
457                 
458                 return percentID;
459
460         }
461         catch(exception& e) {
462                 m->errorOut(e, "Slayer", "snpAB");
463                 exit(1);
464         }
465 }
466 /***********************************************************************/
467 float Slayer::computePercentID(string queryFrag, string parent, int left, int right) {
468         try {
469                 int total = 0;
470                 int matches = 0;
471
472                 for (int i = left; i <= right; i++) {
473                         total++;
474                         if (queryFrag[i] == parent[i]) {
475                                 matches++;
476                         }
477                 }
478
479                 float percentID =( matches/(float)total) * 100;
480                 
481                 return percentID;
482         }
483         catch(exception& e) {
484                 m->errorOut(e, "Slayer", "computePercentID");
485                 exit(1);
486         }
487 }
488 /***********************************************************************/
489 //remove columns that contain any gaps
490 map<int, int> Slayer::verticalFilter(vector<Sequence*> seqs) {
491         try {
492                 //find baseSpots
493                 baseSpots.clear(); 
494                 baseSpots.resize(3);  //query, parentA, parentB
495         
496                 vector<int> gaps;       gaps.resize(seqs[0]->getAligned().length(), 0);
497                 
498                 string filterString = (string(seqs[0]->getAligned().length(), '1'));
499                 
500                 //for each sequence
501                 for (int i = 0; i < seqs.size(); i++) {
502                 
503                         string seqAligned = seqs[i]->getAligned();
504                         
505                         for (int j = 0; j < seqAligned.length(); j++) {
506                                 //if this spot is a gap
507                                 if ((seqAligned[j] == '-') || (seqAligned[j] == '.') || (toupper(seqAligned[j]) == 'N'))        {   gaps[j]++;  }
508                         }
509                 }
510                 
511                 //zero out spot where any sequences have blanks
512                 int numColRemoved = 0;
513                 int count = 0;
514                 map<int, int> maskMap; maskMap.clear();
515
516                 for(int i = 0; i < seqs[0]->getAligned().length(); i++){
517                         if(gaps[i] != 0)        {       filterString[i] = '0';  numColRemoved++;  }
518                         else {
519                                 maskMap[count] = i;
520                                 count++;
521                         }
522                 }
523
524                 //for each sequence
525                 for (int i = 0; i < seqs.size(); i++) {
526                 
527                         string seqAligned = seqs[i]->getAligned();
528                         string newAligned = "";
529                         
530                         int baseCount = 0;
531                         int count = 0;
532                         for (int j = 0; j < seqAligned.length(); j++) {
533                                 //are you a base
534                                 if ((seqAligned[j] != '-') && (seqAligned[j] != '.') && (toupper(seqAligned[j]) != 'N'))        { baseCount++; }
535                         
536                                 //if this spot is not a gap
537                                 if (filterString[j] == '1') { 
538                                         newAligned += seqAligned[j]; 
539                                         baseSpots[i][count] = baseCount;
540                                         count++;
541                                 }
542                         }
543                         
544                         seqs[i]->setAligned(newAligned);
545                 }
546                 
547                 return maskMap;
548         }
549         catch(exception& e) {
550                 m->errorOut(e, "Slayer", "verticalFilter");
551                 exit(1);
552         }
553 }
554 /***********************************************************************/