public void PrepareIndexAndSparseMatrix()
{
var groupedHMM = from thisResult in HMMVerifier.BatchDomainFetcher.Results
group thisResult by thisResult.TargetNamek__BackingField into mygroup
select new { mygroup.Key, Results = mygroup.ToList() };
int counter = 0;
List <SparseMatrixIndexParserV2.Index> tmpIndexes = new List <SparseMatrixIndexParserV2.Index>();
foreach (var groupedResult in groupedHMM)
{
counter++;
SparseMatrixIndexParserV2.Index i = new SparseMatrixIndexParserV2.Index();
i.ID = counter;
i.Name = groupedResult.Key;
i.Description = groupedResult.Results[0].Descriptionk__BackingField + " IDs:" + string.Join(", ", groupedResult.Results.Select(a => a.QNamek__BackingField).Distinct().ToList());
tmpIndexes.Add(i);
}
PatternTools.SparseMatrixIndexParserV2 tmpIndexParser = new SparseMatrixIndexParserV2(tmpIndexes);
tmpIndexParser.SortIndexesByID();
Console.WriteLine("Preparing sparse matrix file");
//First lets extract the classes
myResultPackages.Sort((a, b) => a.ClassLabel.CompareTo(b.ClassLabel));
SparseMatrix tmpSparseMatrix = GenerateDominomicsSparseMatrix(myResultPackages, tmpIndexParser);
//Find out the dims with no values
indexesWithoutValues = tmpIndexParser.TheIndexes.FindAll(a => tmpSparseMatrix.ExtractDimValues(a.ID).Count == 0);
List <PatternTools.SparseMatrixIndexParserV2.Index> cleanIndexes = tmpIndexParser.TheIndexes.Except(indexesWithoutValues).ToList();
for (int i = 0; i < cleanIndexes.Count; i++)
{
cleanIndexes[i].ID = i + 1;
}
indexParserClean = new SparseMatrixIndexParserV2(cleanIndexes);
sparseMatrixClean = GenerateDominomicsSparseMatrix(myResultPackages, indexParserClean);
sparseMatrixClean.ClassDescriptionDictionary = matrixClassDescriptionDictionary;
Console.WriteLine("Done");
}
private SparseMatrix GenerateDominomicsSparseMatrix(List <FileInfoResultPackage> myResultPackages, PatternTools.SparseMatrixIndexParserV2 indexParser)
{
SparseMatrix sm = new SparseMatrix();
//foreach (FileInfoResultPackage thisRP in myResultPackages)
//{
// Console.WriteLine("Processing sparse matrix row for " + thisRP.MyFileInfo.FullName);
// List<int> dims = new List<int>();
// List<double> values = new List<double>();
// foreach (SparseMatrixIndexParserV2.Index index in indexParser.TheIndexes)
// {
// List<HMMResult> hmms = HMMVerifier.BatchDomainFetcher.Results.FindAll(a => a.TargetNamek__BackingField.Equals(index.Name));
// //We can transform this into a concurrent bag and paralelize the loop below
// if (MyInputFormat == InputFormat.SEPro)
// {
// List<PeptideResult> candidatesInDomain = new List<PeptideResult>();
// foreach (HMMResult hmmr in hmms)
// {
// string fastaSeq = locusFastaDict[hmmr.QNamek__BackingField];
// ResultPackage seproRP = (ResultPackage)thisRP.MyResultPackage;
// foreach (PeptideResult pr in seproRP.MyProteins.MyPeptideList)
// {
// MatchCollection mc = Regex.Matches(fastaSeq, PatternTools.pTools.CleanPeptide(pr.CleanedPeptideSequence, true));
// foreach (Match m in mc)
// {
// if (m.Index + m.Length > hmmr.AFromk__BackingField && hmmr.ATok__BackingField > m.Index)
// {
// candidatesInDomain.Add(pr);
// break;
// }
// }
// }
// }
// candidatesInDomain = candidatesInDomain.Distinct().ToList();
// if (candidatesInDomain.Count > 0)
// {
// dims.Add(index.ID);
// values.Add(candidatesInDomain.Sum(a => a.MyScans.Count));
// }
// }
// else if (MyInputFormat == InputFormat.MPex)
// {
// List<AlignmentResult> candidatesInDomain = new List<AlignmentResult>();
// ResultPckg2 pexRP = (ResultPckg2)thisRP.MyResultPackage;
// foreach (AlignmentResult aln in pexRP.Alignments) //Lets cycle through the good alignments
// {
// bool secondBreak = false;
// bool thirdBreak = false;
// foreach (HMMResult hmmr in hmms)
// {
// foreach (Alignment al in aln)
// {
// string pepSeq = Regex.Replace(new string(al.Sequence1), "-", "");
// string fastaSeq = locusFastaDict[hmmr.QNamek__BackingField];
// MatchCollection mc = Regex.Matches(fastaSeq, pepSeq);
// foreach (Match m in mc)
// {
// if (m.Index + m.Length > hmmr.AFromk__BackingField && hmmr.ATok__BackingField > m.Index)
// {
// candidatesInDomain.Add(aln);
// secondBreak = true;
// thirdBreak = true;
// break;
// }
// }
// if (secondBreak)
// {
// break;
// }
// }
// if (thirdBreak)
// {
// break;
// }
// }
// }
// if (candidatesInDomain.Count > 0)
// {
// dims.Add(index.ID);
// values.Add(candidatesInDomain.Count);
// }
// }
// }
// sparseMatrixRow smr = new sparseMatrixRow(thisRP.ClassLabel, dims, values);
// smr.FileName = thisRP.MyFileInfo.FullName;
// sm.addRow(smr);
//}
return(sm);
}
