public static List <ProproteinInterfaceSpreadsheetRecord> MotifSpreadsheetData(List <VectorProteinInterfaceWhole> vectorProteinInterfaceWholeList)
{
if (vectorProteinInterfaceWholeList == null)
{
throw new ArgumentNullException(nameof(vectorProteinInterfaceWholeList));
}
var result = new List <ProproteinInterfaceSpreadsheetRecord>();
// motif by proteinInterface length
var distinctProteinInterfaceLengths = vectorProteinInterfaceWholeList.Select(a => a.ProteinInterfaceLength).Distinct().ToArray();
foreach (var proteinInterfaceLength in distinctProteinInterfaceLengths)
{
for (var index = 0; index < 3; index++)
{
ProproteinInterfaceSpreadsheetRecord record;
if (index == 0)
{
record = Record(vectorProteinInterfaceWholeList.Where(b => b.ProteinInterfaceLength == proteinInterfaceLength).ToList());
}
else if (index == 1)
{
record = Record(vectorProteinInterfaceWholeList.Where(b => b.ProteinInterfaceLength == proteinInterfaceLength && !b.ReversedSequence).ToList());
}
else if (index == 2)
{
record = Record(vectorProteinInterfaceWholeList.Where(b => b.ProteinInterfaceLength == proteinInterfaceLength && b.ReversedSequence).ToList());
}
else
{
throw new IndexOutOfRangeException();
}
if (record == null)
{
continue;
}
record.MotifName = "ProteinInterface Length " + proteinInterfaceLength;
record.MotifSource = "Length";
result.Add(record);
}
}
// motif by common secondary structure pattern
var distinctSecondaryStructures = vectorProteinInterfaceWholeList.Select(a => a.SecondaryStructure).Distinct().ToList();
foreach (var secondaryStructure in distinctSecondaryStructures)
{
for (var index = 0; index < 3; index++)
{
ProproteinInterfaceSpreadsheetRecord record;
if (index == 0)
{
record = Record(vectorProteinInterfaceWholeList.Where(b => b.SecondaryStructure == secondaryStructure).ToList());
}
else if (index == 1)
{
record = Record(vectorProteinInterfaceWholeList.Where(b => b.SecondaryStructure == secondaryStructure && !b.ReversedSequence).ToList());
}
else if (index == 2)
{
record = Record(vectorProteinInterfaceWholeList.Where(b => b.SecondaryStructure == secondaryStructure && b.ReversedSequence).ToList());
}
else
{
throw new IndexOutOfRangeException();
}
if (record == null)
{
continue;
}
record.MotifName = secondaryStructure;
record.MotifSource = "Secondary Structure";
result.Add(record);
}
}
// motif by common interaction vector pattern
for (var vectorType = 0; vectorType < 4; vectorType++)
{
var distinctVectors = vectorProteinInterfaceWholeList.Select(a => a.VectorString(vectorType)).Distinct().ToList();
foreach (var vector in distinctVectors)
{
for (var index = 0; index < 3; index++)
{
ProproteinInterfaceSpreadsheetRecord record;
if (index == 0)
{
record = Record(vectorProteinInterfaceWholeList.Where(b => b.VectorString(vectorType) == vector).ToList());
}
else if (index == 1)
{
record = Record(vectorProteinInterfaceWholeList.Where(b => b.VectorString(vectorType) == vector && !b.ReversedSequence).ToList());
}
else if (index == 2)
{
record = Record(vectorProteinInterfaceWholeList.Where(b => b.VectorString(vectorType) == vector && b.ReversedSequence).ToList());
}
else
{
throw new IndexOutOfRangeException();
}
if (record == null)
{
continue;
}
record.MotifName = vector;
record.MotifSource = VectorProteinInterfaceWhole.VectorStringDescription(vectorType);
result.Add(record);
}
}
}
return(result);
}
public static List <AminoAcidDistributionSpreadsheetRecord> PatternDistributionSpreadsheetRecords(List <VectorProteinInterfaceWhole> vectorProteinInterfaceWholeList, List <string> pdbIdList, List <ISequence> seqList, int vectorType)
{
if (vectorProteinInterfaceWholeList == null)
{
throw new ArgumentNullException(nameof(vectorProteinInterfaceWholeList));
}
if (pdbIdList == null)
{
throw new ArgumentNullException(nameof(pdbIdList));
}
if (seqList == null)
{
throw new ArgumentNullException(nameof(seqList));
}
var result = new List <AminoAcidDistributionSpreadsheetRecord>();
// patterns
var patternProteinDictionary = new Dictionary <string, AminoAcidChainComposition>();
var patternProteinInterfaceDictionary = new Dictionary <string, AminoAcidChainComposition>();
var patternProteinInterfaceInteractionDictionary = new Dictionary <string, AminoAcidChainComposition>();
var patternProteinInterfaceNonInteractionDictionary = new Dictionary <string, AminoAcidChainComposition>();
foreach (var vectorProteinInterfaceWhole in vectorProteinInterfaceWholeList)
{
var interactionBools = vectorProteinInterfaceWhole.InteractionBools();
string pattern;
if (vectorType >= 0 && vectorType <= 3)
{
pattern = vectorProteinInterfaceWhole.VectorString(vectorType);
}
else if (vectorType == 4)
{
pattern = vectorProteinInterfaceWhole.SecondaryStructure;
}
else if (vectorType == 5)
{
pattern = "ProteinInterface Length " + vectorProteinInterfaceWhole.ProteinInterfaceLength;
}
else
{
throw new ArgumentOutOfRangeException(nameof(vectorType));
}
if (!patternProteinDictionary.ContainsKey(pattern))
{
patternProteinDictionary.Add(pattern, new AminoAcidChainComposition());
}
if (!patternProteinInterfaceDictionary.ContainsKey(pattern))
{
patternProteinInterfaceDictionary.Add(pattern, new AminoAcidChainComposition());
}
if (!patternProteinInterfaceInteractionDictionary.ContainsKey(pattern))
{
patternProteinInterfaceInteractionDictionary.Add(pattern, new AminoAcidChainComposition());
}
if (!patternProteinInterfaceNonInteractionDictionary.ContainsKey(pattern))
{
patternProteinInterfaceNonInteractionDictionary.Add(pattern, new AminoAcidChainComposition());
}
var aminoAcids1L = vectorProteinInterfaceWhole.ProteinInterfaceAminoAcids1L();
for (int index = 0; index < aminoAcids1L.Length; index++)
{
var c = aminoAcids1L[index];
patternProteinInterfaceDictionary[pattern].IncrementAminoAcidCount(c);
if (interactionBools[index])
{
patternProteinInterfaceInteractionDictionary[pattern].IncrementAminoAcidCount(c);
}
else
{
patternProteinInterfaceNonInteractionDictionary[pattern].IncrementAminoAcidCount(c);
}
}
patternProteinDictionary[pattern].NumberSamples++;
patternProteinInterfaceDictionary[pattern].NumberSamples++;
patternProteinInterfaceInteractionDictionary[pattern].NumberSamples++;
patternProteinInterfaceNonInteractionDictionary[pattern].NumberSamples++;
var bsSeqList = seqList.Where(a => SequenceIdSplit.SequenceIdToPdbIdAndChainId(a.ID).PdbId == vectorProteinInterfaceWhole.FullProteinInterfaceId.ProteinId && SequenceIdSplit.SequenceIdToPdbIdAndChainId(a.ID).ChainId == SpreadsheetFileHandler.AlphabetLetterRollOver(vectorProteinInterfaceWhole.FullProteinInterfaceId.ChainId)).ToList();
foreach (var chain in bsSeqList)
{
var seq = chain.ConvertToString();
foreach (var c in seq)
{
patternProteinDictionary[pattern].IncrementAminoAcidCount(c);
}
}
}
foreach (var kvp in patternProteinDictionary)
{
var recordAllComposition = new AminoAcidDistributionSpreadsheetRecord();
var recordProteinInterfaceComposition = new AminoAcidDistributionSpreadsheetRecord();
var recordProteinInterfaceInteractionComposition = new AminoAcidDistributionSpreadsheetRecord();
var recordProteinInterfaceNonInteractionComposition = new AminoAcidDistributionSpreadsheetRecord();
result.Add(recordAllComposition);
result.Add(recordProteinInterfaceComposition);
result.Add(recordProteinInterfaceInteractionComposition);
result.Add(recordProteinInterfaceNonInteractionComposition);
recordAllComposition.Pattern = kvp.Key;
recordProteinInterfaceComposition.Pattern = kvp.Key;
recordProteinInterfaceInteractionComposition.Pattern = kvp.Key;
recordProteinInterfaceNonInteractionComposition.Pattern = kvp.Key;
var vectorTypeStr = VectorProteinInterfaceWhole.VectorStringDescription(vectorType);
recordAllComposition.Type = vectorTypeStr;
recordProteinInterfaceComposition.Type = vectorTypeStr;
recordProteinInterfaceInteractionComposition.Type = vectorTypeStr;
recordProteinInterfaceNonInteractionComposition.Type = vectorTypeStr;
recordAllComposition.Metric = "Protein";
recordProteinInterfaceComposition.Metric = "ProteinInterface";
recordProteinInterfaceInteractionComposition.Metric = "ProteinInterface interactions";
recordProteinInterfaceNonInteractionComposition.Metric = "ProteinInterface non interactions";
// number of samples
recordAllComposition.NumberOfSamples = kvp.Value.NumberSamples;
recordProteinInterfaceComposition.NumberOfSamples = patternProteinInterfaceDictionary[kvp.Key].NumberSamples;
recordProteinInterfaceInteractionComposition.NumberOfSamples = patternProteinInterfaceInteractionDictionary[kvp.Key].NumberSamples;
recordProteinInterfaceNonInteractionComposition.NumberOfSamples = patternProteinInterfaceNonInteractionDictionary[kvp.Key].NumberSamples;
// number of amino acids
recordAllComposition.TotalAminoAcids = kvp.Value.AminoAcidGroupsCount[(int)AminoAcidGroups.AminoAcidGroups.EnumAminoAcidGroups.AminoAcids].Sum();
recordProteinInterfaceComposition.TotalAminoAcids = patternProteinInterfaceDictionary[kvp.Key].AminoAcidGroupsCount[(int)AminoAcidGroups.AminoAcidGroups.EnumAminoAcidGroups.AminoAcids].Sum();
recordProteinInterfaceInteractionComposition.TotalAminoAcids = patternProteinInterfaceInteractionDictionary[kvp.Key].AminoAcidGroupsCount[(int)AminoAcidGroups.AminoAcidGroups.EnumAminoAcidGroups.AminoAcids].Sum();
recordProteinInterfaceNonInteractionComposition.TotalAminoAcids = patternProteinInterfaceNonInteractionDictionary[kvp.Key].AminoAcidGroupsCount[(int)AminoAcidGroups.AminoAcidGroups.EnumAminoAcidGroups.AminoAcids].Sum();
var allPercentage = AminoAcidChainComposition.ConvertToPercentage(kvp.Value);
var proteinInterfacePercentage = AminoAcidChainComposition.ConvertToPercentage(patternProteinInterfaceDictionary[kvp.Key]);
var proteinInterfaceInteractionPercentage = AminoAcidChainComposition.ConvertToPercentage(patternProteinInterfaceInteractionDictionary[kvp.Key]);
var proteinInterfaceNonInteractionPercentage = AminoAcidChainComposition.ConvertToPercentage(patternProteinInterfaceNonInteractionDictionary[kvp.Key]);
foreach (AminoAcidGroups.AminoAcidGroups.EnumAminoAcidGroups enumAminoAcidGroups in Enum.GetValues(typeof(AminoAcidGroups.AminoAcidGroups.EnumAminoAcidGroups)))
{
var groupItemsTotal = AminoAcidGroups.AminoAcidGroups.GetTotalSubgroups(enumAminoAcidGroups);
for (var groupItemIndex = 0; groupItemIndex < groupItemsTotal; groupItemIndex++)
{
recordAllComposition.Distribution[(int)enumAminoAcidGroups][groupItemIndex] = allPercentage.AminoAcidGroupsCount[(int)enumAminoAcidGroups][groupItemIndex];
recordProteinInterfaceComposition.Distribution[(int)enumAminoAcidGroups][groupItemIndex] = proteinInterfacePercentage.AminoAcidGroupsCount[(int)enumAminoAcidGroups][groupItemIndex];
recordProteinInterfaceInteractionComposition.Distribution[(int)enumAminoAcidGroups][groupItemIndex] = proteinInterfaceInteractionPercentage.AminoAcidGroupsCount[(int)enumAminoAcidGroups][groupItemIndex];
recordProteinInterfaceNonInteractionComposition.Distribution[(int)enumAminoAcidGroups][groupItemIndex] = proteinInterfaceNonInteractionPercentage.AminoAcidGroupsCount[(int)enumAminoAcidGroups][groupItemIndex];
}
}
}
return(result);
}
