internal EpitopeLearningData Subtract(EpitopeLearningData other)
{
EpitopeLearningData rgOut = new EpitopeLearningData(string.Format("{0}-{1}", Name, other.Name));
foreach (Pair <string, Hla> aMerAndHlaToLength in Keys)
{
if (!other.ContainsKey(aMerAndHlaToLength))
{
rgOut[aMerAndHlaToLength] = this[aMerAndHlaToLength];
}
}
return(rgOut);
}
//internal static EpitopeLearningData GetSyfpeithiCollection(HlaResolution hlaResolution, KmerDefinition kmerDefinition)
//{
// Debug.Assert(kmerDefinition.FullMerCount == kmerDefinition.EpitopeMerCount); //!!!this is all we have code for
// Debug.Assert(8 <= kmerDefinition.EpitopeMerCount && kmerDefinition.EpitopeMerCount <= 11);
// Debug.Assert(hlaResolution is ABMixedHlaResolution); //!!!this is all we have code for
// EpitopeLearningData rg = new EpitopeLearningData("Syfpeithi");
// string fileInput = @"W_2005-04-05_09_57_55_syfpeithiClean.txt";//!!!const
// using (StreamReader streamreader = Predictor.OpenResource(fileInput))
// {
// string line = streamreader.ReadLine();
// Debug.Assert(line == "Epitope ID Epitope Sequence HLA kMerLength Anchor 1 Anchor 2 Anchor 3");
// while (null != (line = streamreader.ReadLine()))
// {
// string[] fieldCollection = line.Split('\t');
// SpecialFunctions.CheckCondition(fieldCollection.Length >= 4); //!!!raise error
// string aaSequence = fieldCollection[1].Trim();
// if (aaSequence.Length != kmerDefinition.FullMerCount)
// {
// continue;
// }
// if (aaSequence.Contains("X"))
// {
// continue;
// }
// string hlaAsString = fieldCollection[2].Trim();
// HlaToLength hlaToLength = HlaToLength.GetInstance(hlaAsString, hlaResolution);
// MerAndHlaToLength aMerAndHlaToLength = MerAndHlaToLength.GetInstance(aaSequence, hlaToLength, kmerDefinition);
// rg[aMerAndHlaToLength] = true;
// }
// }
// return rg;
//}
//internal static EpitopeLearningData GetMhcPepCollection(HlaResolution hlaResolution, KmerDefinition kmerDefinition)
//{
// Debug.Fail("The MHC Pep data needs it's A68's expanded before it can be used");
// Debug.Assert(kmerDefinition.EpitopeMerCount == 9 && kmerDefinition.FullMerCount == 9); //!!!this is the case we have code for
// EpitopeLearningData rg = new EpitopeLearningData("MhcPep");
// string fileInput = @"W_2005-04-05_11_43_49_MHCPEPcleanDuplicatesRemoved.txt"; //!!!const
// using (StreamReader streamreader = Predictor.OpenResource(fileInput))
// {
// string line = streamreader.ReadLine();
// Debug.Assert(line == "Epitopes Original HLA (4 digit where possible) Our Canonical HLA Length");
// while (null != (line = streamreader.ReadLine()))
// {
// string[] fieldCollection = line.Split('\t');
// SpecialFunctions.CheckCondition(fieldCollection.Length == 4); //!!!raise error
// string aaSequence = fieldCollection[0].Trim();
// if (aaSequence.Length != kmerDefinition.FullMerCount)
// {
// continue;
// }
// string hlaAsString = fieldCollection[2].Trim();
// if (hlaAsString == "B15" || hlaAsString == "A68")
// {
// Debug.WriteLine("Skipping 2-digit B15 and A68's");
// continue;
// }
// HlaToLength hlaToLength = HlaToLength.GetInstanceABMixed(hlaAsString);
// int length = int.Parse(fieldCollection[3]);
// SpecialFunctions.CheckCondition(length == aaSequence.Length); //!!!raise error
// MerAndHlaToLength aMerAndHlaToLength = MerAndHlaToLength.GetInstance(aaSequence, hlaToLength, kmerDefinition);
// rg[aMerAndHlaToLength] = true;
// }
// }
// return rg;
//}
public static EpitopeLearningData Union(params EpitopeLearningData[] rgrg)
{
EpitopeLearningData rgOut = new EpitopeLearningData(null);
StringBuilder sb = new StringBuilder();
foreach (EpitopeLearningData rgIn in rgrg)
{
if (sb.Length > 0)
{
sb.Append('+');
}
sb.Append(rgIn.Name);
foreach (KeyValuePair <Pair <string, Hla>, bool> aMerAndHlaToLengthWithLabel in rgIn)
{
rgOut[aMerAndHlaToLengthWithLabel.Key] = aMerAndHlaToLengthWithLabel.Value;
}
}
rgOut.Name = sb.ToString();
return(rgOut);
}
//internal EpitopeLearningData[] Split(int cCrossValPart, Random aRandom)
//{
// List<KeyValuePair<MerAndHlaToLength, bool>> shuffleList = new List<KeyValuePair<MerAndHlaToLength, bool>>();
// foreach (KeyValuePair<Pair<string, Hla>, bool> merAndHlaToLengthWithLabel in this)
// {
// shuffleList.Add(merAndHlaToLengthWithLabel);
// int iRandomPos = aRandom.Next(shuffleList.Count);
// shuffleList[shuffleList.Count - 1] = shuffleList[iRandomPos];
// shuffleList[iRandomPos] = merAndHlaToLengthWithLabel;
// }
// EpitopeLearningData[] rgrg = new EpitopeLearningData[cCrossValPart];
// for (int irgrg = 0; irgrg < rgrg.Length; ++irgrg)
// {
// rgrg[irgrg] = new EpitopeLearningData(string.Format("{0}{1}", Name, irgrg));
// }
// for (int iShuffleList = 0; iShuffleList < shuffleList.Count; ++iShuffleList)
// {
// KeyValuePair<MerAndHlaToLength, bool> merAndHlaToLengthWithLabel = shuffleList[iShuffleList];
// int iSet = iShuffleList * cCrossValPart / shuffleList.Count;
// rgrg[iSet].Add(merAndHlaToLengthWithLabel.Key, merAndHlaToLengthWithLabel.Value);
// }
// return rgrg;
//}
public static EpitopeLearningData GetDbWhole(HlaFactory hlaFactory, int eLength, string datasetName, string fileOrResourceName)
{
Set <string> wantedSet = CreateSourceSet(datasetName);
EpitopeLearningData rg = new EpitopeLearningData(datasetName);
//SpecialFunctions.CheckCondition(hlaResolution.Equals(HlaResolution.ABMixed));
foreach (Dictionary <string, string> row in Predictor.TabFileTable(fileOrResourceName, "peptide hla source label", false))
{
string peptide = row["peptide"];
SpecialFunctions.CheckCondition(Biology.GetInstance().LegalPeptide(peptide), string.Format("Peptide, '{0}', contains illegal char.", peptide));
if (peptide.Length != eLength) //!!!const
{
continue;
}
string source = row["source"];
Set <string> providedSet = CreateSourceSet(source);
//Debug.Assert(providedSet.IsSubsetOf(Set<string>.GetInstance(new string[] { "Aplus", "LANL", "IEDB" }))); // real assert
if (providedSet.IntersectionIsEmpty(wantedSet))
{
continue;
}
Hla hla = hlaFactory.GetGroundInstance(row["hla"]);
//HlaToLength hlaToLength = HlaToLength.GetInstance(hla, hlaResolution);
Pair <string, Hla> peptideAndHla = new Pair <string, Hla>(peptide, hla);
//MerAndHlaToLength aMerAndHlaToLength = MerAndHlaToLength.GetInstance(peptide, hlaToLength, kmerDefinition);
string label = row["label"];
SpecialFunctions.CheckCondition(label == "0" || label == "1", string.Format("Warning: Epitope example {0} has unknown label {1} and will be ignored.", peptideAndHla, label));
rg[peptideAndHla] = (label == "1");
}
return(rg);
}
internal EpitopeLearningData Union(EpitopeLearningData other)
{
return(Union(this, other));
}