public void SearchRare(ArrayList rareBlockList)
{
for (int s = 0; s < snpCount - 2; s++)
{
ArrayList group = new ArrayList();
int end = s;
foreach (int i in this.individual)
{
if (this.snp[i, s] * this.snp[i, s + 1] * this.snp[i, s + 2] > 0)
{
group.Add(i);
}
}
if (group.Count >= 3)
{
// Console.Write("{0} {1}\r\n", s, group.Count);
end += 2;
Boolean flag = true;
while (flag)
{
if (end == this.snpCount - 1)
{
break;
}
foreach (int j in group)
{
if (this.snp[j, end + 1] == 0)
{
//end++;
flag = false;
break;
}
}
end++;
}
RareBlock newRareBlock = new RareBlock();
newRareBlock.SetStart(s);
newRareBlock.SetEnd(end - 1);
newRareBlock.InitialSequence(newRareBlock.end - newRareBlock.start + 1);
for (int i = 0; i < newRareBlock.sequence.Length; i++)
{
newRareBlock.SetSequence(i, 1);
}
newRareBlock.group = (int[])group.ToArray(typeof(int));
RareBlock lastRareBlock = new RareBlock();
rareBlockList.Add(newRareBlock);
/*
* if (newRareBlock.group.Length < this.individual.Length * 100 / 100)
* {
* if (rareBlockList.Count < 1)
* rareBlockList.Add(newRareBlock);
* else
* {
* lastRareBlock = (RareBlock)rareBlockList[rareBlockList.Count - 1];
* //lastRareBlock2 = (RareBlock)rareBlockList[rareBlockList.Count - 2];
* if (lastRareBlock.end == newRareBlock.end && lastRareBlock.group.Length == newRareBlock.group.Length)
* {
* int i = 0;
* int j = 0;
* for (i = 0; i < newRareBlock.group.Length; i++)
* {
* if (lastRareBlock.group[j] == newRareBlock.group[i])
* j++;
* else break;
* }
* if (i != newRareBlock.group.Length)
* rareBlockList.Add(newRareBlock);
*
* }
* else
* rareBlockList.Add(newRareBlock);
* }
* }
*/
}
}
}
//private void MarcovModel(double p,int start, int end, int[] haplotype, int[] linkageData)
public bool Pvalue(double[,] linkageData, ArrayList pairwiseIBDList, int index)
{
this.InitialP();
this.InitialHaplotype(this.GetSequence().Length);
this.SetHaplotype(this.GetSequence());
this.SetP(0, 0);
for (int i = 0; i < this.GetP().Length; i++)
{
if (i == 0) //first snp of the region, set the shared haplotype to be the major allele.
{
if (this.GetSequence(i) > 1)
{
this.SetHaplotype(i, 2);
}
else
{
this.SetHaplotype(i, 0);
}
this.SetP(0, 1);
continue;
}
if (this.haplotype[i] != 1) //if the haplotype is determined (0 for share common, 2 for share rare, 1 for uncertain)
{
int currentHap = this.GetHaplotype(i);
int lastHap = this.GetHaplotype(i - 1);
if (lastHap == 1)
{
this.SetP(i, 1);
continue;
}
if (currentHap == 0) //for cases to check, major->major, major->minor, minor->major,minor->minor;
{
if (lastHap == 0)
{
this.SetP(i, linkageData[this.GetStart() + i - 1, 2]);
}
else
{
this.SetP(i, linkageData[this.GetStart() + i - 1, 4]);
}
}
else
{
if (lastHap == 0)
{
this.SetP(i, linkageData[this.GetStart() + i - 1, 3]);
}
else
{
this.SetP(i, linkageData[this.GetStart() + i - 1, 5]);
}
}
}
else //if the haplotype at position i is not determined (the value is 1), but in this case,
// the previous snp haplotype can determine this haplotype (say, last one is major and major->minor is 0%, then determine that this position is major)
{
this.SetP(i, 1);
}
}
//*** now we should check the break point of the haplotype. (at which position the conversion P is almost zero.(which is simulated to be 1-power(0.95, sampleSize)))
int lastBreak = 0; //last break point position.
bool flag = false; //false means that no break point. true means there is at least one break point.
//int lastBreak = this.GetSequence().Length;
//return true;
for (int i = 0; i < this.GetP().Length; i++)
{
if (this.GetP(i) <= 0.001)//1.0 / (linkageData.GetUpperBound(0) + 1))
{
if (i - lastBreak < 100)
{
lastBreak = i;
}
else
{
RareBlock newBlock = new RareBlock();
newBlock.InitialSequence(i - lastBreak);
for (int j = 0; j < newBlock.GetSequence().Length; j++)
{
newBlock.SetSequence(j, this.GetSequence(j + lastBreak));
}
newBlock.SetStart(lastBreak);
newBlock.SetEnd(i - 1);
newBlock.SetGroup(this.GetGroup());
pairwiseIBDList.Insert(index + 1, newBlock);
lastBreak = i;
}
flag = true;
}
}
if (flag)
{
pairwiseIBDList.RemoveAt(index);
return(false);
}
return(true);
}
public bool PvalueWithMM(double[,] linkageData, ArrayList pairwiseIBDList, int index)
{
this.InitialP();
this.InitialHaplotype(this.GetSequence().Length);
this.SetHaplotype(this.GetSequence());
this.SetP(0, 0);
for (int i = 0; i < this.GetP().Length; i++)
{
if (i == 0) //first snp of the region, set the shared haplotype to be the major allele.
{
if (this.GetSequence(i) > 1)
{
this.SetHaplotype(i, 2);
}
else
{
this.SetHaplotype(i, 0);
}
this.SetP(0, 1);
continue;
}
if (this.haplotype[i] != 1) //if the haplotype is determined (0 for share common, 2 for share rare, 1 for uncertain)
{
int currentHap = this.GetHaplotype(i);
int lastHap = this.GetHaplotype(i - 1);
if (lastHap == 1)
{
this.SetP(i, 1);
continue;
}
if (currentHap == 0) //for cases to check, major->major, major->minor, minor->major,minor->minor;
{
if (lastHap == 0)
{
this.SetP(i, linkageData[this.GetStart() + i - 1, 2]);
}
else
{
this.SetP(i, linkageData[this.GetStart() + i - 1, 4]);
}
}
else
{
if (lastHap == 0)
{
this.SetP(i, linkageData[this.GetStart() + i - 1, 3]);
}
else
{
this.SetP(i, linkageData[this.GetStart() + i - 1, 5]);
}
}
}
else //if the haplotype at position i is not determined (the value is 1), but in this case,
// the previous snp haplotype can determine this haplotype (say, last one is major and major->minor is 0%, then determine that this position is major)
{
if (this.haplotype[i - 1] == 0)
{
if (linkageData[this.GetStart() + i - 1, 2] == (double)1)
{
this.SetSequence(i, 0);
this.SetHaplotype(i, 0);
this.SetP(i, 1);
continue;
}
else if (linkageData[this.GetStart() + i - 1, 3] == (double)1)
{
this.SetSequence(i, 2);
this.SetHaplotype(i, 2);
this.SetP(i, 1);
continue;
}
}
else
{
if (linkageData[this.start + i - 1, 4] == (double)1)
{
this.SetSequence(i, 0);
this.SetHaplotype(i, 0);
this.SetP(i, 1);
continue;
}
else if (linkageData[this.start + i - 1, 5] == (double)1)
{
this.SetSequence(i, 2);
this.SetHaplotype(i, 2);
this.SetP(i, 1);
continue;
}
}
if (i == this.GetP().Length - 1)
{
if (this.GetHaplotype(i - 1) == 0)
{
if (linkageData[this.GetStart() + i - 1, 2] >= 0.5)
{
this.SetHaplotype(i, 0);
this.SetP(i, linkageData[this.GetStart() + i - 1, 2]);
}
else
{
this.SetHaplotype(i, 2);
this.SetP(i, linkageData[this.GetStart() + i - 1, 3]);
}
}
else
{
if (linkageData[this.GetStart() + i - 1, 4] >= 0.5)
{
this.SetHaplotype(i, 0);
this.SetP(i, linkageData[this.GetStart() + i - 1, 4]);
}
else
{
this.SetHaplotype(i, 2);
this.SetP(i, linkageData[this.GetStart() + i - 1, 5]);
}
}
continue;
}
//if still here, means that we should decide the haplotype by p value.
int j = i; //j is the start positon of uncertain haplotype region
while (i < this.p.Length - 1)
{
if (this.haplotype[i + 1] == 1)
{
//this.p[i + 1] = 1;
i++;
}
else
{
i++;
break;
}
//determine the length of the uncertain haplotype region.
}
int length = i - j; //the length, now i is the end of the region+1
/*
* if (length >= 3) // if the region is >=3, it should be the rareblock which we start the searching.
* {
* for (int k = 0; k < length; k++)
* this.haplotype[j+k] = 2;
* i =j -1; //go back to the start position of the uncertain region.
* continue;
* }
*/
//this will cause many impossible combination, omitted now. 7th,Nov,2008
//if not, we should calculate which haplotype is most likely to be shared by these individuals.
double[,] markovP = new double[length + 1, 4]; //marcov chain data, for each row stores:
//[,0]:major of this snp will choose major(0) or minor(2), [,1]the p value of this chosen
//[,2]:major of this snp will choose major(0) or minor(2), [,3]the p value of this chosen
markovP[length, 0] = haplotype[i];
markovP[length, 2] = haplotype[i];
if (i < this.GetP().Length - 1)
{
if (markovP[length, 0] == 0)
{
markovP[length, 1] = linkageData[j + length - 1, 2];
markovP[length, 3] = linkageData[j + length - 1, 4];
}
else
{
markovP[length, 1] = linkageData[j + length - 1, 3];
markovP[length, 3] = linkageData[j + length - 1, 5];
}
}//fill the last row first, coz which one should follow is determined at the end of the uncertain region.
else
{
markovP[length, 1] = 0.5;
markovP[length, 3] = 0.5;
}
for (int m = length - 1; m >= 0; m--) //then fill the row from bottom to top.
{
double AB = linkageData[j + m - 1 + this.start, 2] * markovP[m + 1, 1];
double Ab = linkageData[j + m - 1 + this.start, 3] * markovP[m + 1, 3];
if (AB >= Ab)
{
markovP[m, 0] = 0;
markovP[m, 1] = AB;
}
else
{
markovP[m, 0] = 2;
markovP[m, 1] = Ab;
}
double aB = linkageData[j + m - 1 + this.start, 4] * markovP[m + 1, 1];
double ab = linkageData[j + m - 1 + this.start, 5] * markovP[m + 1, 3];
if (aB >= ab)
{
markovP[m, 2] = 0;
markovP[m, 3] = aB;
}
else
{
markovP[m, 2] = 2;
markovP[m, 3] = ab;
}
}// finished filling the marcov table.
/* if (this.sequence[j - 1] == 0)
* {
* this.p[i] = (int)markovP[0, 1];
* }
* else
* {
* this.p[i] = (int)markovP[0, 3];
* }
*/
int n = 0;
while (n < length) //read from the marcov table and assign all the haplotype with shared region.
{
if (this.haplotype[j - 1 + n] == 0)
{
this.SetHaplotype(j + n, (int)markovP[n, 0]);
}
else
{
this.SetHaplotype(j + n, (int)markovP[n, 2]);
}
n++;
}
i = j - 1; //go back the start position of the uncertain region and caculate the p value (now the uncertain region is certain).
}
}
//*** now we should check the break point of the haplotype. (at which position the conversion P is almost zero.(which is simulated to be 1-power(0.95, sampleSize)))
int lastBreak = 0; //last break point position.
bool flag = false; //false means that no break point. true means there is at least one break point.
//int lastBreak = this.GetSequence().Length;
//return true;
for (int i = 0; i < this.GetP().Length; i++)
{
if (this.GetP(i) <= 0.001)//1.0 / (linkageData.GetUpperBound(0) + 1))
{
if (i - lastBreak < 100)
{
lastBreak = i;
}
else
{
RareBlock newBlock = new RareBlock();
newBlock.InitialSequence(i - lastBreak);
for (int j = 0; j < newBlock.GetSequence().Length; j++)
{
newBlock.SetSequence(j, this.GetSequence(j + lastBreak));
}
newBlock.SetStart(lastBreak);
newBlock.SetEnd(i - 1);
newBlock.SetGroup(this.GetGroup());
pairwiseIBDList.Insert(index + 1, newBlock);
lastBreak = i;
}
flag = true;
}
}
if (flag)
{
pairwiseIBDList.RemoveAt(index);
return(false);
}
return(true);
}
public void Combine(int i, ArrayList rareBlockList, ArrayList combinedBlockList, SnpDataSet dataSet)
{
if (i == rareBlockList.Count - 1)
{
return; //reach the end of the rareBlockList.
}
ArrayList newGroup = new ArrayList();
ArrayList remainGroup = new ArrayList();
bool flag = true;
for (int j = i + 1; j < i + 50 && j < rareBlockList.Count; j++) // find combined block from this block to 100 blocks away.
{
RareBlock block = new RareBlock();
//Console.WriteLine("{0} {1}", i, j);
block = (RareBlock)((RareBlock)rareBlockList[j]).MemberwiseClone();
if (block.GetStart() < this.GetEnd())
{
continue;
}
int m = 0;
int n = 0;
newGroup.Clear();
remainGroup.Clear();
for (m = 0; m < this.GetGroup().Length; m++) //find the intersection of the two group of individuals.
{
if (n == block.GetGroup().Length)
{
remainGroup.Add(this.GetGroup(m));
break;
}
if (this.group[m] == block.GetGroup(n))
{
newGroup.Add(this.group[m]);
n++;
}
else if (this.group[m] > block.GetGroup(n))
{
n++;
m--;
}
else
{
remainGroup.Add(this.GetGroup(m));
}
}
if (newGroup.Count > 2) //more than 1 guys appare in the two rare blocks
{
//Console.Write("{0} {1} {2}\t", i, j, newGroup.Count);
//foreach (int ind in newGroup)
// Console.Write("{0} ",ind);
//Console.WriteLine();
//Console.ReadKey();
RareBlock newRareBlock = new RareBlock();
newRareBlock.SetStart(this.GetStart());
newRareBlock.SetEnd(block.GetEnd());
newRareBlock.SetGroup((int[])newGroup.ToArray(typeof(int)));
if (newRareBlock.CheckCommon(dataSet)) //check the snp data between the rare blocks
//if (true)
{
if (newRareBlock.GetGroup().Length == 3)
{
newRareBlock.Extend(dataSet);
if (newRareBlock.GetEnd() - newRareBlock.GetStart() > 170)
{
if (combinedBlockList.Count > 0)
{
RareBlock lastRareBlock = (RareBlock)((RareBlock)combinedBlockList[combinedBlockList.Count - 1]).MemberwiseClone();
if (!lastRareBlock.CheckSame(this))
{
if (this.CheckContain(lastRareBlock))
{
combinedBlockList[combinedBlockList.Count - 1] = this.GetCopy();
}
else
{
combinedBlockList.Add(this.GetCopy());
}
}
}
else
{
combinedBlockList.Add(this.GetCopy());
}
}
}
else
{
newRareBlock.Combine(j, rareBlockList, combinedBlockList, dataSet);//if pass, continue combining.
}
//break; //When we find another block to combine to this block ,stop searching for this block.
//Time saving a lot, can miss some, discuss later on.
if (this.group.Length - newRareBlock.group.Length == 0)
{
flag = false;
break;
}
//else if (newRareBlock.group.Length / (this.group.Length - newRareBlock.group.Length) >= 3) break;
}
/*
* if (this.group.Length - newRareBlock.group.Length >= 3)
* {
* RareBlock remainGroupBlock = new RareBlock();
* remainGroupBlock.start = this.start;
* remainGroupBlock.end = this.end;
* remainGroupBlock.sequence = this.sequence;
* remainGroupBlock.group = (int[])remainGroup.ToArray(typeof(int));
* remainGroupBlock.Combine(j, rareBlockList, combinedBlockList, dataSet);
* }
*/
}
}
if (flag && (this.GetEnd() - this.GetEnd()) > 70)
{
if (combinedBlockList.Count > 0)
{
RareBlock lastRareBlock = (RareBlock)combinedBlockList[combinedBlockList.Count - 1];
if (!lastRareBlock.CheckSame(this))
{
if (this.CheckContain(lastRareBlock))
{
combinedBlockList[combinedBlockList.Count - 1] = this.GetCopy();
}
else
{
combinedBlockList.Add(this.GetCopy());
}
}
}
else
{
combinedBlockList.Add(this.GetCopy());
}
}
}