/*
* string s의 모든 로테이션을 구하고, 그것들로 matrix를 구성한다음,
* 그 matrxix의 row들을 lexically하게 정렬한 후, 각 row들의 마지막 문자들을 취한 것이 BWT이다.
*
*/
static private BWTransformation BWTransform(string s)
{
int l = s.Length;
char[,] ar = AllRotations(s);
List <SortBuffer> sl = new List <SortBuffer>();
string[] sa = new string[l];
sa = Useful.Char2DArrToStringArr(ar, l, l);
for (int i = 0; i < l; i++)
{
sl.Add(new SortBuffer(sa[i], i));
}
char[] r = new char[l];
int[] rps = new int[l];
//List<string> sl = sa.ToList<string>();
StringBuilder f = new StringBuilder();
IEnumerable <SortBuffer> query = sl.OrderBy(sb => sb.s);
sl = new List <SortBuffer>();
foreach (SortBuffer e in query)
{
sl.Add(e);
}
//Making BWM is done.
for (int i = 0; i < l; i++)
{
r[i] = sl[i].s[l - 1]; f.Append(sl[i].s[0]); rps[i] = sl[i].rp;
}
BWTransformation tr = new BWTransformation();
tr.L = Useful.CharToString(r, l);
tr.F = f.ToString();
tr.inputidx = sl[0].rp;
tr.rps = Useful.ArrayCopy <int>(rps, l);
return(tr);
}
public static GibbsMotifData ForStringSeq(string[] Seqs, int L, int W)//W is motif length and L is the length of one sequence.
{
int N = Seqs.Length;
int[] setA = new int[N]; //set of starting indices
int pa; //previous a.
int seq1idx;
string seq1;
WeightMatrixForGibbs <char> wmat;
Random rnd = new Random();
for (int i = 0; i < N; i++)
{
setA[i] = rnd.Next(L - W + 1);
}
do
{
seq1idx = rnd.Next(N);
seq1 = Seqs[seq1idx];
wmat = MakeWeightMatrixForStringSeq(Seqs, N, W, seq1idx, setA);
pa = setA[seq1idx];
setA[seq1idx] = wmat.ArgMaxScore(seq1.ToArray <char>(), N, L);
Console.WriteLine(Useful.CharToString(wmat.MotifMax(), W));
Console.WriteLine(CutString(seq1, setA[seq1idx], setA[seq1idx] + W - 1));
Console.ReadLine();
} while ((setA[seq1idx] != pa) || (Useful.CharToString(wmat.MotifMax(), W) != CutString(seq1, setA[seq1idx], setA[seq1idx] + W - 1)));
//(Useful.CharToString(wmat.MotifMax(),W) != CutString(seq1, setA[seq1idx], setA[seq1idx]+W-1))
//위 조건은 좀 위험한게, shungke처럼 정확한 motif가 있으면 상관없는데, weak motif일경우 조건을 절대로 만족할 수 없게 될지도...
StringBuilder motif = new StringBuilder();
for (int i = 0; i < W; i++)
{
motif.Append(seq1[pa + i]);
}
return(new GibbsMotifData(motif.ToString(), wmat, Seqs, N, L, W));
}