public void TestReadFromFile()
{
var counts = new GeneCountTableFormat().ReadFromFile(@"../../../data/genetable.tsv");
Assert.AreEqual(2, counts.GeneValues.Count);
Assert.AreEqual(4, counts.Samples.Length);
Assert.AreEqual("ENSG00000000003", counts.GeneValues[0][0]);
Assert.AreEqual("TSPAN6", counts.GeneValues[0][1]);
Assert.AreEqual("IG-001", counts.Samples[0]);
Assert.AreEqual(128, counts.Count[0, 0]);
Assert.AreEqual(60, counts.Count[0, 1]);
Assert.AreEqual(288, counts.Count[0, 2]);
Assert.AreEqual(9, counts.Count[0, 3]);
}
public GeneCountTable CalculateFPKM(out double[] sampleCounts, out double[] geneLengths)
{
Progress.SetMessage("Reading gene length from {0} ...", options.GeneLengthFile);
var columnNames = FileUtils.ReadColumnNames(options.GeneLengthFile);
var lengthIndex = columnNames.ToList().FindIndex(m => m.ToLower().Equals("length"));
if (lengthIndex < 0)
{
throw new Exception("Cannot find length column in file " + options.GeneLengthFile);
}
var geneLengthMap = new MapItemReader(0, lengthIndex).ReadFromFile(options.GeneLengthFile).ToDictionary(m => m.Key, m => double.Parse(m.Value.Value));
Progress.SetMessage("Reading count table from {0} ...", options.InputFile);
var counts = new GeneCountTableFormat().ReadFromFile(options.InputFile);
if (!string.IsNullOrEmpty(options.KeyRegex))
{
var reg = new Regex(options.KeyRegex);
geneLengthMap = geneLengthMap.ToDictionary(l => reg.Match(l.Key).Groups[1].Value, l => l.Value);
counts.GeneValues[0][0] = reg.Match(counts.GeneValues[0][0]).Groups[1].Value;
}
Dictionary<string, double> sampleReads;
if (File.Exists(options.SampleReadsFile))
{
Progress.SetMessage("Reading sample reads from {0} ...", options.SampleReadsFile);
sampleReads = new MapItemReader(0, 1).ReadFromFile(options.SampleReadsFile).ToDictionary(m => m.Key, m => double.Parse(m.Value.Value));
}
else //use total mapped reads as total reads
{
sampleReads = new Dictionary<string, double>();
for (int iSample = 0; iSample < counts.Samples.Length; iSample++)
{
double itotal = 0.0;
for (int iGene = 0; iGene < counts.GeneValues.Count; iGene++)
{
itotal += counts.Count[iGene, iSample];
}
sampleReads[counts.Samples[iSample]] = itotal;
}
}
foreach (var sample in counts.Samples)
{
if (!sampleReads.ContainsKey(sample))
{
throw new Exception(string.Format("No sample {0} found at sample reads file {1}", sample, options.SampleReadsFile));
}
}
foreach (var geneValues in counts.GeneValues)
{
if (!geneLengthMap.ContainsKey(geneValues[0]))
{
throw new Exception(string.Format("No gene {0} found at gene length file {1}", geneValues[0], options.GeneLengthFile));
}
}
sampleCounts = (from sample in counts.Samples
select sampleReads[sample]).ToArray();
geneLengths = (from geneValues in counts.GeneValues
select geneLengthMap[geneValues[0]]).ToArray();
for (int iGene = 0; iGene < geneLengths.Length; iGene++)
{
for (int iSample = 0; iSample < sampleCounts.Length; iSample++)
{
counts.Count[iGene, iSample] = counts.Count[iGene, iSample] * 1000000000 / (geneLengths[iGene] * sampleCounts[iSample]);
}
}
return counts;
}
