public void HarmonizePolarities(GenotypeDictionary dict1, GenotypeDictionary dict2, out GenotypeDictionary harmonizedDict1, out GenotypeDictionary harmonizedDict2)
{
//Harmonizes dict2 after dict1 and removes all entries from A/T or G/C experiments and incompatible
//SNP types and items from SNPs which are monomorphic in both source.
string tempExp, tempAlleles1, tempAlleles2;
CIStringCollection uniqueExps;
Dictionary <string, string> allelesDict1, allelesDict2;
int counter;
//Find all experiments and their SNP types (the experiment names should be the same in both dictionaries, use dict1 here).
uniqueExps = new CIStringCollection();
allelesDict1 = new Dictionary <string, string>(StringComparer.InvariantCultureIgnoreCase);
allelesDict2 = new Dictionary <string, string>(StringComparer.InvariantCultureIgnoreCase);
OnStatusChange("Extracting experiments and SNP types");
counter = 0;
foreach (string tempKey in dict1.Keys)
{
tempExp = dict1.GetExperiment(tempKey);
tempAlleles1 = dict1[tempKey].ToUpper();
tempAlleles2 = dict2[tempKey].ToUpper();
if (!uniqueExps.Contains(tempExp))
{
//This experiment has not been processed before.
uniqueExps.Add(tempExp);
//Get the unique alleles from dictionary 1.
if (tempAlleles1.Substring(0, 1) != tempAlleles1.Substring(2, 1))
{
//The alleles are different, add both.
allelesDict1.Add(tempExp, tempAlleles1.Substring(0, 1) + tempAlleles1.Substring(2, 1));
}
else
{
//The alleles are the same, add only the first.
allelesDict1.Add(tempExp, tempAlleles1.Substring(0, 1));
}
//Get the unique alleles from dictionary 2.
if (tempAlleles2.Substring(0, 1) != tempAlleles2.Substring(2, 1))
{
//The alleles are different, add both.
allelesDict2.Add(tempExp, tempAlleles2.Substring(0, 1) + tempAlleles2.Substring(2, 1));
}
else
{
//The alleles are the same, add only the first.
allelesDict2.Add(tempExp, tempAlleles2.Substring(0, 1));
}
}
else
{
//This experiment has been processed before.
//Add the first allele from dictionary one if it does not already exist in the string.
if (!allelesDict1[tempExp].Contains(tempAlleles1.Substring(0, 1)))
{
allelesDict1[tempExp] = allelesDict1[tempExp] + tempAlleles1.Substring(0, 1);
}
//Add the second allele from dictionary one if it does not already exist in the string.
if (!allelesDict1[tempExp].Contains(tempAlleles1.Substring(2, 1)))
{
allelesDict1[tempExp] = allelesDict1[tempExp] + tempAlleles1.Substring(2, 1);
}
//Add the first allele from dictionary two if it does not already exist in the string.
if (!allelesDict2[tempExp].Contains(tempAlleles2.Substring(0, 1)))
{
allelesDict2[tempExp] = allelesDict2[tempExp] + tempAlleles2.Substring(0, 1);
}
//Add the second allele from dictionary one if it does not already exist in the string.
if (!allelesDict2[tempExp].Contains(tempAlleles2.Substring(2, 1)))
{
allelesDict2[tempExp] = allelesDict2[tempExp] + tempAlleles2.Substring(2, 1);
}
}
counter++;
if (counter % 100 == 0)
{
OnStatusChange("Extracting experiments and SNP types (processed genotype " + counter + ")");
}
}
//Remove data from SNPs with more than two alleles in source 1 from both dictionaries and save.
MyRemovedExpTooManyAlleles1 = new CIStringCollection();
OnStatusChange("Extracting harmonizable data step 1/6");
counter = 0;
foreach (string exp in allelesDict1.Keys)
{
if (allelesDict1[exp].Length > 2)
{
dict1.RemoveExperiment(exp);
dict2.RemoveExperiment(exp);
MyRemovedExpTooManyAlleles1.Add(exp);
}
counter++;
if (counter % 100 == 0)
{
OnStatusChange("Extracting harmonizable data step 1/6 (processed experiment " + counter + ")");
}
}
//Remove data from SNPs with more than two alleles in source 2 from both dictionaries and save.
MyRemovedExpTooManyAlleles2 = new CIStringCollection();
OnStatusChange("Extracting harmonizable data step 2/6");
counter = 0;
foreach (string exp in allelesDict2.Keys)
{
if (allelesDict2[exp].Length > 2)
{
dict1.RemoveExperiment(exp);
dict2.RemoveExperiment(exp);
MyRemovedExpTooManyAlleles2.Add(exp);
}
counter++;
if (counter % 100 == 0)
{
OnStatusChange("Extracting harmonizable data step 2/6 (processed experiment " + counter + ")");
}
}
//Remove data from SNPs which are monomorphic in both sources and save.
MyRemovedExpMonomorphicInBoth = new CIStringCollection();
OnStatusChange("Extracting harmonizable data step 3/6");
counter = 0;
foreach (string exp in allelesDict1.Keys)
{
if (allelesDict1[exp].Length == 1 && allelesDict2[exp].Length == 1)
{
dict1.RemoveExperiment(exp);
dict2.RemoveExperiment(exp);
MyRemovedExpMonomorphicInBoth.Add(exp);
}
counter++;
if (counter % 100 == 0)
{
OnStatusChange("Extracting harmonizable data step 3/6 (processed experiment " + counter + ")");
}
}
//Remove data from SNPs with incompatible SNP types from both dictionaries and save.
MyRemovedExpIncompatibleSNPTypes = new CIStringCollection();
OnStatusChange("Extracting harmonizable data step 4/6");
counter = 0;
foreach (string exp in allelesDict1.Keys)
{
if (!MyRemovedExpTooManyAlleles1.Contains(exp) && !MyRemovedExpTooManyAlleles2.Contains(exp) &&
!MyRemovedExpMonomorphicInBoth.Contains(exp))
{
if (!this.IsCompatibleSNPTypes(allelesDict1[exp], allelesDict2[exp]))
{
dict1.RemoveExperiment(exp);
dict2.RemoveExperiment(exp);
MyRemovedExpIncompatibleSNPTypes.Add(exp);
}
}
counter++;
if (counter % 100 == 0)
{
OnStatusChange("Extracting harmonizable data step 4/6 (processed experiment " + counter + ")");
}
}
//Remove data from SNPs with A/T or G/C SNPs from both dictionaries and save.
MyRemovedExpATCG = new CIStringCollection();
OnStatusChange("Extracting harmonizable data step 5/6");
counter = 0;
foreach (string exp in allelesDict1.Keys)
{
if (!MyRemovedExpTooManyAlleles1.Contains(exp) && !MyRemovedExpTooManyAlleles2.Contains(exp) &&
!MyRemovedExpMonomorphicInBoth.Contains(exp) && !MyRemovedExpIncompatibleSNPTypes.Contains(exp))
{
if (allelesDict1[exp].Length > 1)
{
if (this.GetNumberOfATalleles(allelesDict1[exp]) != 1)
{
//Either 0 or 2 A or T alleles.
dict1.RemoveExperiment(exp);
dict2.RemoveExperiment(exp);
MyRemovedExpATCG.Add(exp);
}
}
else
{
if (this.GetNumberOfATalleles(allelesDict2[exp]) != 1)
{
dict1.RemoveExperiment(exp);
dict2.RemoveExperiment(exp);
MyRemovedExpATCG.Add(exp);
}
}
}
counter++;
if (counter % 100 == 0)
{
OnStatusChange("Extracting harmonizable data step 5/6 (processed experiment " + counter + ")");
}
}
//Store only data from experiments passing all tests.
harmonizedDict1 = new GenotypeDictionary();
harmonizedDict2 = new GenotypeDictionary();
OnStatusChange("Extracting harmonizable data step 6/6");
counter = 0;
foreach (string tempKey in dict1.Keys)
{
tempExp = dict1.GetExperiment(tempKey);
if (!MyRemovedExpTooManyAlleles1.Contains(tempExp) &&
!MyRemovedExpTooManyAlleles2.Contains(tempExp) &&
!MyRemovedExpMonomorphicInBoth.Contains(tempExp) &&
!MyRemovedExpIncompatibleSNPTypes.Contains(tempExp) &&
!MyRemovedExpATCG.Contains(tempExp))
{
harmonizedDict1.Add(tempKey, dict1[tempKey]);
harmonizedDict2.Add(tempKey, dict2[tempKey]);
}
counter++;
if (counter % 100 == 0)
{
OnStatusChange("Extracting harmonizable data step 6/6 (processed genotype " + counter + ")");
}
}
//Go through dict1. If the SNP type for the current row is different than
//the SNP type in dict2, convert the row in dict2.
OnStatusChange("Harmonizing");
counter = 0;
foreach (string tempKey in harmonizedDict1.Keys)
{
tempExp = harmonizedDict1.GetExperiment(tempKey);
tempAlleles1 = allelesDict1[tempExp];
tempAlleles2 = allelesDict2[tempExp];
if (tempAlleles1.Length == 2 && tempAlleles2.Length == 2)
{
if ((
(tempAlleles1.Substring(0, 1) == tempAlleles2.Substring(0, 1)) &&
(tempAlleles1.Substring(1, 1) == tempAlleles2.Substring(1, 1))
) ||
(
(tempAlleles1.Substring(0, 1) == tempAlleles2.Substring(1, 1)) &&
(tempAlleles1.Substring(1, 1) == tempAlleles2.Substring(0, 1))
)
)
{
//They are the same.
}
else
{
harmonizedDict2[tempKey] = this.ReverseComplement(harmonizedDict2[tempKey]);
}
}
else if (tempAlleles1.Length == 1 && tempAlleles2.Length == 2)
{
if (tempAlleles1 != tempAlleles2.Substring(0, 1) && tempAlleles1 != tempAlleles2.Substring(1, 1))
{
harmonizedDict2[tempKey] = this.ReverseComplement(harmonizedDict2[tempKey]);
}
}
else if (tempAlleles1.Length == 2 && tempAlleles2.Length == 1)
{
if (tempAlleles2 != tempAlleles1.Substring(0, 1) && tempAlleles2 != tempAlleles1.Substring(1, 1))
{
harmonizedDict2[tempKey] = this.ReverseComplement(harmonizedDict2[tempKey]);
}
}
else
{
throw new Exception("Inappropriate number of alleles detected when converting SNP types");
}
counter++;
if (counter % 100 == 0)
{
OnStatusChange("Harmonizing (processed genotype " + counter + ")");
}
}
}
public void Compare()
{
GenotypeDictionary unionDict1, unionDict2;
GenotypeDictionary harmonizedDict1, harmonizedDict2;
string tempValue1, tempValue2;
int counter;
MyMissingIn1 = new GenotypeDictionary();
MyMissingIn2 = new GenotypeDictionary();
MyCompared = new GenotypeResultPairCollection();
MyIdentical = new GenotypeResultPairCollection();
MyDifferent = new GenotypeResultPairCollection();
unionDict1 = new GenotypeDictionary();
unionDict2 = new GenotypeDictionary();
MyHarmonizer = new ComparisonHarmonizer();
//Sort out keys from results1 which are also in results2.
counter = 0;
OnStatusChange("Find common genotypes, step 1");
foreach (string key1 in MySourceInfo1.ValidGenotypes.Keys)
{
if (MySourceInfo2.ValidGenotypes.Contains(key1))
{
unionDict1.Add(key1, MySourceInfo1.ValidGenotypes[key1]);
}
else
{
MyMissingIn2.Add(key1);
}
counter++;
if (counter % 100 == 0)
{
OnStatusChange("Find common genotypes, step 1 (processed genotype " + counter + ")");
}
}
//Sort out keys from results2 which are also in results1.
counter = 0;
OnStatusChange("Find common genotypes, step 2");
foreach (string key2 in MySourceInfo2.ValidGenotypes.Keys)
{
if (MySourceInfo1.ValidGenotypes.Contains(key2))
{
unionDict2.Add(key2, MySourceInfo2.ValidGenotypes[key2]);
}
else
{
MyMissingIn1.Add(key2);
}
counter++;
if (counter % 100 == 0)
{
OnStatusChange("Find common genotypes, step 2 (processed genotype " + counter + ")");
}
}
//Make sure the two union dictionaries are of the same length.
if (unionDict1.Count != unionDict2.Count)
{
throw new Exception("Comparison error. Lists are of unequal length.");
}
//Harmonize polarities if the settings say so.
if (MySettings.Mode == InvestigationMode.Genotype && MySettings.HarmonizePolarities)
{
OnStatusChange("Harmonizing polarities");
MyHarmonizer.MyStatusChangeHandler += new InvestigationStatusChangeHandler(MyHarmonizer_StatusChanged);
MyHarmonizer.HarmonizePolarities(unionDict1, unionDict2, out harmonizedDict1, out harmonizedDict2);
unionDict1 = harmonizedDict1;
unionDict2 = harmonizedDict2;
}
//Perform comparison.
counter = 0;
OnStatusChange("Comparing genotypes");
foreach (string key in unionDict1.Keys)
{
tempValue1 = unionDict1[key];
tempValue2 = unionDict2[key];
if (this.IsEqual(tempValue1, tempValue2))
{
MyIdentical.Add(key, MySourceInfo1.ValidGenotypes[key], MySourceInfo2.ValidGenotypes[key]);
}
else
{
MyDifferent.Add(key, MySourceInfo1.ValidGenotypes[key], MySourceInfo2.ValidGenotypes[key]);
}
MyCompared.Add(key, MySourceInfo1.ValidGenotypes[key], MySourceInfo2.ValidGenotypes[key]);
counter++;
if (counter % 100 == 0)
{
OnStatusChange("Comparing genotypes (processed genotype " + counter + ")");
}
}
}
public void Load(DataTable results, InvestigationMode cmpType)
{
string tempItem, tempExp, tempAlleles;
GenotypeDictionary tempDuplFailures, tempGenotypeDict, tempDuplTestDict;
GenotypeCollection tempNoResults;
GenotypeResultCollection tempInvalidAlleles;
MyTotalGenotypesCount = results.Rows.Count;
//Initiate dictionaries and collections.
tempGenotypeDict = new GenotypeDictionary();
tempDuplTestDict = new GenotypeDictionary();
tempDuplFailures = new GenotypeDictionary();
tempInvalidAlleles = new GenotypeResultCollection();
tempNoResults = new GenotypeCollection();
//First perform duplicate test.
foreach (DataRow row in results.Rows)
{
//Read values.
tempItem = row["Item"].ToString();
tempExp = row["Experiment"].ToString();
tempAlleles = row["Alleles"].ToString();
if (tempAlleles.ToUpper() != MyMissingValueCode.ToUpper())
{
if (this.IsValidAlleleCombination(tempAlleles, cmpType))
{
if (tempDuplTestDict.Contains(tempItem, tempExp))
{
//This key already exists, check if the alleles are different.
if (tempDuplTestDict[tempItem, tempExp].ToUpper() != tempAlleles.ToUpper())
{
//Remember this key as having a duplicate failure.
if (!tempDuplFailures.Contains(tempItem, tempExp))
{
tempDuplFailures.Add(tempItem, tempExp, "");
}
}
}
else
{
//Did not exist already.
tempDuplTestDict.Add(tempItem, tempExp, tempAlleles);
}
}
}
}
//Now go through the values again to avoid those with duplicate failures.
foreach (DataRow row in results.Rows)
{
//Read values.
tempItem = row["Item"].ToString();
tempExp = row["Experiment"].ToString();
tempAlleles = row["Alleles"].ToString();
if (tempAlleles.ToUpper() == MyMissingValueCode.ToUpper())
{
//Skipped because of no result.
tempNoResults.Add(tempItem, tempExp);
}
else if (!this.IsValidAlleleCombination(tempAlleles, cmpType))
{
//Skipped because invalid alleles.
tempInvalidAlleles.Add(tempItem, tempExp, tempAlleles);
}
else if (!tempDuplFailures.Contains(tempItem, tempExp) && !tempGenotypeDict.Contains(tempItem, tempExp))
{
//Neither a duplicate failure nor already existing, go ahead and add.
tempGenotypeDict.Add(tempItem, tempExp, tempAlleles);
}
}
MyValidGenotypes = tempGenotypeDict;
MyDuplFailures = tempDuplFailures;
MyNoResults = tempNoResults;
MyInvalidAlleles = tempInvalidAlleles;
}