public override IEnumerable <string> Process()
{
using (var file = new PlinkBedRandomFile(_options.InputFile)
{
Progress = this.Progress
})
{
var locusList = file.Data.Locus;
var individualList = file.Data.Individual;
Progress.SetRange(0, locusList.Count);
for (int i = 0; i < locusList.Count; i++)
{
Progress.SetPosition(i);
var locus = locusList[i];
var data = file.Read(locus.MarkerId);
int count1 = 0;
int count2 = 0;
int validSample = 0;
for (int j = 0; j < individualList.Count; j++)
{
if (PlinkData.IsMissing(data[0, j], data[1, j]))
{
continue;
}
validSample++;
if (data[0, j])
{
count2++;
}
else
{
count1++;
}
if (data[1, j])
{
count2++;
}
else
{
count1++;
}
}
locus.Allele1Frequency = ((double)(count1)) / (count1 + count2);
locus.TotalSample = individualList.Count;
locus.ValidSample = validSample;
}
PlinkLocus.WriteToFile(_options.OutputFile, locusList, false, true, true);
}
return(new string[] { _options.OutputFile });
}
private static void Validate(PlinkData data)
{
Assert.AreEqual(2, data.Locus.Count);
Assert.AreEqual(7, data.Individual.Count);
Assert.AreEqual("GCCCGC0", data.LocusAllele1(0));
Assert.AreEqual("GGGCGG0", data.LocusAllele2(0));
Assert.AreEqual("ATTTA0T", data.LocusAllele1(1));
Assert.AreEqual("ATAAA0A", data.LocusAllele2(1));
Assert.AreEqual("2,1,1,0,2,1,3", data.LocusGenoType(0, ","));
Assert.AreEqual("2,0,1,1,2,3,1", data.LocusGenoType(1, ","));
}
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;
}
public override IEnumerable <string> Process()
{
var result = new List <string>();
var bimfile = Path.ChangeExtension(options.InputFile, ".bim");
var snps = PlinkLocus.ReadFromBimFile(bimfile, false, false);
snps.RemoveAll(m => IsIndel(m) || IsMissing(m));
var snpItems = (from snp in snps
select new SNPItem()
{
Chrom = snp.Chromosome,
Name = snp.MarkerId,
Position = snp.PhysicalPosition,
Allele1 = snp.Allele1[0],
Allele2 = snp.Allele2
}).ToList();
var nameMap = snpItems.FillDbsnpIdByPosition(options.DbsnpFile, this.Progress);
using (var sw = new StreamWriter(options.OutputPrefix + ".namemap"))
{
sw.WriteLine("NewName\tOldName");
foreach (var n in nameMap)
{
sw.WriteLine("{0}\t{1}", n.Key, n.Value);
}
}
//remove all snps without corresponding dbsnp entry
snpItems.RemoveAll(m => m.DbsnpRefAllele == ' ');
var nameDic = snpItems.ToGroupDictionary(m => m.Name);
foreach (var n in nameDic)
{
if (n.Value.Count > 1)
{
Console.Error.WriteLine("Duplicated SNP:" + n.Key);
foreach (var v in n.Value)
{
Console.Error.WriteLine("{0}:{1}-{2}:{3},{4}:{5},{6}", n.Key, v.Chrom, v.Position, v.Allele1, v.Allele2, v.DbsnpRefAllele, v.DbsnpAltAllele);
}
}
}
if (File.Exists(options.G1000File))
{
snpItems.FindAllele2FrequencyFrom1000GomeByName(options.G1000File, this.Progress);
}
if (File.Exists(options.FastaFile))
{
snpItems.FillReferenceAlleleFromFasta(options.FastaFile, this.Progress);
}
Dictionary <string, StrandAction> actionMap = new Dictionary <string, StrandAction>();
var statFile = options.OutputPrefix + ".stat";
result.Add(statFile);
using (var sw = new StreamWriter(statFile))
{
sw.WriteLine("Name\tChromosome\tPosition\tSource_Allele1\tSource_Allele2\tReference_Allele\tDbsnp_RefAllele\tDbsnp_AltAllele\tDbsnp_IsReversed\tG1000_RefAllele\tG1000_AltAllele\tG1000_MAF\tAction");
foreach (var v in snpItems)
{
StrandAction action = v.SuggestAction();
sw.WriteLine("{0}\t{1}\t{2}\t{3}\t{4}\t{5}\t{6}\t{7}\t{8}\t{9}\t{10}\t{11:0.####}\t{12}", v.Name, v.Chrom, v.Position, v.Allele1, v.Allele2, v.RefChar, v.DbsnpRefAllele, v.DbsnpAltAllele, v.DbsnpIsReversed, v.G1000Allele1, v.G1000Allele2, v.G1000Allele2Frequency, action);
actionMap[v.Name] = action;
}
}
using (var reader = new PlinkBedRandomFile(options.InputFile)
{
Progress = this.Progress
})
{
var data = reader.Data;
var chrs = (from v in snpItems select v.Chrom).Distinct().OrderBy(m => m).ToArray();
foreach (var chr in chrs)
{
var genfile = string.Format("{0}.{1}.gen", options.OutputPrefix, chr.ToString().PadLeft(2, '0'));
result.Add(genfile);
var map = FileUtils.ChangeExtension(genfile, ".sample");
new GwasSampleFormat().WriteToFile(map, data.Individual);
//save gen file
using (var sw = new StreamWriter(genfile))
{
sw.NewLine = Environment.NewLine;
var chrItems = snpItems.Where(m => m.Chrom == chr).ToList();
GenomeUtils.SortChromosome(chrItems, m => chr.ToString(), m => m.Position);
foreach (var snp in chrItems)
{
var ldata = reader.Read(nameMap[snp.Name]);
var action = actionMap[snp.Name];
sw.Write("{0} {1} {2} {3} {4}", snp.Chrom, snp.Name, snp.Position, snp.DbsnpRefAllele, snp.DbsnpAltAllele);
for (int individualIndex = 0; individualIndex < data.Individual.Count; individualIndex++)
{
if (PlinkData.IsMissing(ldata[0, individualIndex], ldata[1, individualIndex]))
{
sw.Write(" 0 0 0");
}
else
{
char alle1, alle2;
if (StrandAction.Switch == action || StrandAction.FlipSwitch == action)
{
alle1 = ldata[0, individualIndex] ? snp.DbsnpAltAllele : snp.DbsnpRefAllele;
alle2 = ldata[1, individualIndex] ? snp.DbsnpAltAllele : snp.DbsnpRefAllele;
}
else
{
alle1 = ldata[0, individualIndex] ? snp.DbsnpRefAllele : snp.DbsnpAltAllele;
alle2 = ldata[1, individualIndex] ? snp.DbsnpRefAllele : snp.DbsnpAltAllele;
}
if (alle1 != alle2)
{
sw.Write(" 0 1 0");
}
else if (alle1 == snp.DbsnpRefAllele)
{
sw.Write(" 1 0 0");
}
else
{
sw.Write(" 0 0 1");
}
}
}
sw.WriteLine();
}
}
}
}
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);
}
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;
}
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;
}
