Example Program
Motif Finder
Examples for how to start a motif search using SeqAn's Motif Finder.
File "find_motif.cpp"
A tutorial about motif search.
1#include <iostream>
2#include "seqan/find_motif.h"
3
4using namespace seqan;
5
Function to output found motifs.
6template <typename TMotifFinder>
7void printMotifs(TMotifFinder & finder)
8{
9    for (int i = 0; i < (int) motifCount(finder); ++i)
10    {
11        std::cout << i << ": " << getMotif(finder, i) << std::endl;
12    }
13}
14
15int main() 
16{
17    std::srand((unsigned) time(NULL));
18
Motif search on a small set of nucleotide sequences.
19    unsigned int t = 3;        //number of input sequences
20    unsigned int n = 6;        //length of sequence
21    unsigned int l = 4;        //length of motif
22    unsigned int d = 1;        //number of substitutions
23    bool is_exact = true;    //size of Hamming distance
24    unsigned int h = 0;        //size of the neighborhood considering at first
25
26    String<DnaString> dataset;
27    appendValue(dataset,DnaString("ACAGCA"));
28    appendValue(dataset,DnaString("AGGCAG"));
29    appendValue(dataset,DnaString("TCAGTC"));
30    
Application of ePatternBranching (h=0)
31    MotifFinder<Dna, EPatternBranching> finder_epb1(t,l,d,is_exact,h);
32    findMotif(finder_epb1,dataset,Omops());
33    std::cout << getMotif(finder_epb1) << std::endl;
34
Application of ePatternBranching (h=0)
35    MotifFinder<Dna, EPatternBranching> finder_epb2(t,l,d,is_exact,h);
36    findMotif(finder_epb2,dataset,Oops());
37    std::cout << getMotif(finder_epb2) << std::endl;
38
Application of Pms1-Zoops
39    MotifFinder<Dna, Pms1> finder_pms1(l,d,is_exact);
40    findMotif(finder_pms1,dataset,Zoops());
41    printMotifs(finder_pms1); 
42
Application of Pmsp-Tcm
43    MotifFinder<Dna, Pmsp> finder_pmsp(l,d,is_exact);
44    findMotif(finder_pmsp,dataset,Tcm());
45    printMotifs(finder_pmsp); 
46    
Application of PROJECTION-Oops
47    unsigned int m = t*(n-l+1);
48    MotifFinder<Dna, Projection> finder_proj(t,l,m,d,is_exact);
49    findMotif(finder_proj, dataset, Oops());
50    printMotifs(finder_proj);
51
Application of PROJECTION-Omops
52    MotifFinder<Dna, Projection> finder_proj_omops(t,l,m,d,is_exact);
53    findMotif(finder_proj_omops, dataset, Omops());
54    printMotifs(finder_proj_omops);
55
Application of PROJECTION-Zoops
56    MotifFinder<Dna, Projection> finder_proj_zoops(t,l,m,d,is_exact);
57    findMotif(finder_proj_zoops, dataset, Zoops());
58    printMotifs(finder_proj_zoops);
59    
Application of PROJECTION-Tcm
60    MotifFinder<Dna, Projection> finder_proj_tcm(t,l,m,d,is_exact);
61    findMotif(finder_proj_tcm, dataset, Tcm());
62    printMotifs(finder_proj_tcm);
63
64    return 0;
65}
66
Output
[0]: AGCC

[0]: AGCC

[0]: AAGC 
[1]: ACAG
[2]: AGAC
[3]: AGCC
[4]: AGGA
[5]: AGTA
[6]: CAGA
[7]: CAGG
[8]: CCAG
[9]: CGCA
[10]: CGGC
[11]: GCAG
[12]: TCAG
[13]: TGCA

[0]: AAGC
[1]: AAGT
[2]: AATC
[3]: ACAG
[4]: ACGC
[5]: ACTC
[6]: AGAC
[7]: AGCA
[8]: AGCC
[9]: AGGA
[10]: AGGC
[11]: AGGG
[12]: AGGT
[13]: AGTA
[14]: AGTC
[15]: AGTG
[16]: AGTT
[17]: ATGC
[18]: ATTC
[19]: CAAT
[20]: CACT
[21]: CAGA
[22]: CAGC
[23]: CAGG
[24]: CATT
[25]: CCAG
[26]: CCGT
[27]: CGCA
[28]: CGGC
[29]: CGGT
[30]: CGTC
[31]: CTGT
[32]: GAAG
[33]: GACA
[34]: GAGT
[35]: GCAA
[36]: GCAC
[37]: GCAG
[38]: GCAT
[39]: GCCA
[40]: GCCG
[41]: GCGG
[42]: GCTG
[43]: GGAA
[44]: GGAG
[45]: GGCC
[46]: GGCG
[47]: GGCT
[48]: GGGA
[49]: GGGC
[50]: GGTA
[51]: GGTC
[52]: GTAG
[53]: GTCA
[54]: TAAG
[55]: TAGT
[56]: TCAA
[57]: TCAC
[58]: TCAG
[59]: TCAT
[60]: TCCG
[61]: TCGG
[62]: TCTG
[63]: TGAG
[64]: TGCA
[65]: TGGC
[66]: TGTC
[67]: TTAG

AGCC

AGCC

TCAG

TCAG
SeqAn - Sequence Analysis Library - www.seqan.de
 

Page built @2013/07/11 09:12:35