/// <summary>
/// Read individual from fam file of bed format, or from ped file of ped format
/// </summary>
/// <param name="fileName">fam/ped file</param>
/// <returns>list of PlinkIndividual</returns>
public static List<PlinkIndividual> ReadFromFile(string fileName)
{
var result = new List<PlinkIndividual>();
using (var sr = new StreamReader(fileName))
{
string line;
var comms = new[] { '\t', ' ' };
while ((line = sr.ReadLine()) != null)
{
line = line.Trim();
if (string.IsNullOrEmpty(line))
{
continue;
}
var parts = line.Split(comms);
if (string.IsNullOrEmpty(parts[1]))
{
continue;
}
var ind = new PlinkIndividual();
ind.Fid = parts[0];
ind.Iid = parts[1];
ind.Pat = parts[2];
ind.Mat = parts[3];
ind.Sexcode = parts[4];
ind.Phenotype = double.Parse(parts[5]);
result.Add(ind);
}
}
return result;
}
/// <summary>
/// Read individual from fam file of bed format, or from ped file of ped format
/// </summary>
/// <param name="fileName">fam/ped file</param>
/// <returns>list of PlinkIndividual</returns>
public static List <PlinkIndividual> ReadFromFile(string fileName)
{
var result = new List <PlinkIndividual>();
using (var sr = new StreamReader(fileName))
{
string line;
var comms = new[] { '\t', ' ' };
while ((line = sr.ReadLine()) != null)
{
line = line.Trim();
if (string.IsNullOrEmpty(line))
{
continue;
}
var parts = line.Split(comms);
if (string.IsNullOrEmpty(parts[1]))
{
continue;
}
var ind = new PlinkIndividual();
ind.Fid = parts[0];
ind.Iid = parts[1];
ind.Pat = parts[2];
ind.Mat = parts[3];
ind.Sexcode = parts[4];
ind.Phenotype = double.Parse(parts[5]);
result.Add(ind);
}
}
return(result);
}
public PlinkData ReadFromFile(string fileName)
{
var famFile = FileUtils.ChangeExtension(fileName, ".fam");
if (!File.Exists(famFile))
{
throw new FileNotFoundException("File not found: " + famFile);
}
var bimFile = FileUtils.ChangeExtension(fileName, ".bim");
if (!File.Exists(bimFile))
{
throw new FileNotFoundException("File not found: " + bimFile);
}
var result = new PlinkData();
result.Individual = PlinkIndividual.ReadFromFile(famFile);
result.Locus = PlinkLocus.ReadFromBimFile(bimFile);
result.AllocateDataMemory();
OpenBinaryFile(fileName);
try
{
if (IsSNPMajor)
{
for (int i = 0; i < result.Locus.Count; i++)
{
int j = 0;
while (j < result.Individual.Count)
{
var b = ReadByte();
int c = 0;
while (c < 7 && j < result.Individual.Count)
{
result.IsHaplotype1Allele2[i, j] = b[c++];
result.IsHaplotype2Allele2[i, j] = b[c++];
j++;
}
}
}
}
else
{
for (int i = 0; i < result.Individual.Count; i++)
{
int j = 0;
while (j < result.Locus.Count)
{
var b = ReadByte();
int c = 0;
while (c < 7 && j < result.Locus.Count)
{
result.IsHaplotype1Allele2[j, i] = b[c++];
result.IsHaplotype2Allele2[j, i] = b[c++];
j++;
}
}
}
}
}
finally
{
_reader.Close();
_reader = null;
}
return(result);
}
private PlinkData ReadFromFileWithoutIndel(string fileName)
{
var result = ReadLocus(fileName);
result.Individual = PlinkIndividual.ReadFromFile(fileName);
result.AllocateDataMemory();
var allele1 = new char[result.Locus.Count, result.Individual.Count];
var allele2 = new char[result.Locus.Count, result.Individual.Count];
int individual = -1;
//reading data
using (var sr = new StreamReader(fileName))
{
string line;
while ((line = sr.ReadLine()) != null)
{
individual++;
var parts = line.Split(' ');
for (int snp = 0; snp < result.Locus.Count; snp++)
{
var locus = result.Locus[snp];
var pos = 6 + snp * 2;
var a1 = parts[pos];
var a2 = parts[pos + 1];
allele1[snp, individual] = a1[0];
allele2[snp, individual] = a2[0];
}
}
}
bool alleleAssigned = HasAlleleAssigned(result);
for (int locus = 0; locus < result.Locus.Count; locus++)
{
if (!alleleAssigned)
{
var count = new Dictionary <char, int>();
bool bFound = false;
for (int ind = 0; ind < result.Individual.Count; ind++)
{
var a1 = allele1[locus, ind];
if (a1 == PlinkLocus.MISSING_CHAR)
{
continue;
}
var a2 = allele2[locus, ind];
if (a2 == PlinkLocus.MISSING_CHAR)
{
continue;
}
if (a1 != a2)
{
result.Locus[locus].AlleleChar1 = a1;
result.Locus[locus].AlleleChar2 = a2;
bFound = true;
break;
}
int v;
if (count.TryGetValue(a1, out v))
{
count[a1] = v + 1;
}
else
{
count[a1] = 1;
}
}
if (!bFound)
{
var orderedCount = count.ToList().OrderByDescending(m => m.Value).ToList();
if (orderedCount.Count == 0)
{
continue;
}
if (orderedCount.Count == 1)
{
result.Locus[locus].AlleleChar1 = orderedCount[0].Key;
result.Locus[locus].AlleleChar2 = orderedCount[0].Key;
continue;
}
if (orderedCount.Count == 2)
{
result.Locus[locus].AlleleChar1 = orderedCount[0].Key;
result.Locus[locus].AlleleChar2 = orderedCount[1].Key;
continue;
}
throw new Exception(string.Format("There are more than 3 alleles for locus {0} : {1}", result.Locus[locus].MarkerId, (from c in orderedCount select c.Key.ToString()).Merge(", ")));
}
result.Locus.ForEach(m =>
{
m.Allele1 = m.AlleleChar1.ToString();
m.Allele2 = m.AlleleChar2.ToString();
});
}
else
{
result.Locus.ForEach(m =>
{
m.AlleleChar1 = m.Allele1[0];
m.AlleleChar2 = m.Allele2[0];
});
}
var l1 = result.Locus[locus].AlleleChar1;
//assign value
for (int ind = 0; ind < result.Individual.Count; ind++)
{
var a1 = allele1[locus, ind];
var a2 = allele2[locus, ind];
if (a1 == PlinkLocus.MISSING_CHAR || a2 == PlinkLocus.MISSING_CHAR)
{
result.IsHaplotype1Allele2[locus, ind] = true;
result.IsHaplotype2Allele2[locus, ind] = false;
continue;
}
result.IsHaplotype1Allele2[locus, ind] = a1 != l1;
result.IsHaplotype2Allele2[locus, ind] = a2 != l1;
}
}
allele1 = null;
allele2 = null;
return(result);
}
public void OpenBinaryFile(string fileName)
{
var famFile = FileUtils.ChangeExtension(fileName, ".fam");
if (!File.Exists(famFile))
{
throw new FileNotFoundException("File not found: " + famFile);
}
var bimFile = FileUtils.ChangeExtension(fileName, ".bim");
if (!File.Exists(bimFile))
{
throw new FileNotFoundException("File not found: " + bimFile);
}
Data = new PlinkData();
Data.Individual = PlinkIndividual.ReadFromFile(famFile);
Data.Locus = PlinkLocus.ReadFromBimFile(bimFile);
//Data.Locus.ForEach(m => m.MarkerId = m.MarkerId.ToLower());
Data.BuildMap();
DoOpenFile(fileName);
BitArray b = ReadByte();
bool v1_bfile = true;
if ((b[2] && b[3] && b[5] && b[6]) && !(b[0] || b[1] || b[4] || b[7]))
{
// Next number
b = ReadByte();
if ((b[0] && b[1] && b[3] && b[4]) && !(b[2] || b[5] || b[6] || b[7]))
{
b = ReadByte();
IsSNPMajor = b[0];
if (IsSNPMajor)
{
Progress.SetMessage("Detected that binary PED file is v1.00 SNP-major mode\n");
}
else
{
Progress.SetMessage("Detected that binary PED file is v1.00 individual-major mode\n");
}
}
else
{
v1_bfile = false;
}
}
else
{
v1_bfile = false;
}
// Reset file if < v1
if (!v1_bfile)
{
Progress.SetMessage("Warning, old BED file <v1.00 : will try to recover...\n");
DoOpenFile(fileName);
b = ReadByte();
}
// If 0.99 file format
if ((!v1_bfile) && (b[1] || b[2] || b[3] || b[4] || b[5] || b[6] || b[7]))
{
Progress.SetMessage(" *** Possible problem: guessing that BED is < v0.99 *** ");
Progress.SetMessage(" *** High chance of data corruption, spurious results *** ");
IsSNPMajor = false;
DoOpenFile(fileName);
}
else if (!v1_bfile)
{
IsSNPMajor = b[0];
Progress.SetMessage("Binary PED file is v0.99\n");
if (IsSNPMajor)
{
Progress.SetMessage("Detected that binary PED file is in SNP-major mode\n");
}
else
{
Progress.SetMessage("Detected that binary PED file is in individual-major mode\n");
}
}
_startPosition = _reader.BaseStream.Position;
}
