public static Dictionary <string, MotifCounter> MotifDistinctWithCount(List <ProproteinInterfaceSpreadsheetRecord> proproteinInterfaceSpreadsheetRecordList)
{
if (proproteinInterfaceSpreadsheetRecordList == null)
{
throw new ArgumentNullException(nameof(proproteinInterfaceSpreadsheetRecordList));
}
var result = new Dictionary <string, MotifCounter>();
foreach (var record in proproteinInterfaceSpreadsheetRecordList)
{
var motifs = record.Motifs().Distinct().ToArray();
foreach (var motif in motifs)
{
if (!result.ContainsKey(motif))
{
var motifCounter = new MotifCounter()
{
Motif = motif,
MotifTooGeneral = ProproteinInterfaceMotif.IsMotifTooGeneral(motif),
TotalFwd = 0,
TotalRev = 0,
TotalMix = 0,
};
result.Add(motif, motifCounter);
}
if (record.Direction == "Fwd")
{
result[motif].TotalFwd++;
}
else if (record.Direction == "Rev")
{
result[motif].TotalRev++;
}
else if (record.Direction == "Mix")
{
result[motif].TotalMix++;
}
}
}
return(result);
}
public static ProproteinInterfaceSpreadsheetRecord Record(List <VectorProteinInterfaceWhole> vectorProteinInterfaceWholeList)
{
if (vectorProteinInterfaceWholeList == null || vectorProteinInterfaceWholeList.Count == 0)
{
return(null);
}
var proteinInterfaceSequenceList = vectorProteinInterfaceWholeList.Select(a => a.ProteinInterfaceAminoAcids1L()).ToList();
var proteinInterfaceInteractionSequenceList = vectorProteinInterfaceWholeList.Select(a => string.Join("", a.ProteinInterfaceAminoAcids1L().Select((b, i) => a.InteractionBools()[i] ? b : ' ').ToList())).ToList();
var proteinInterfaceNonInteractionSequenceList = vectorProteinInterfaceWholeList.Select(a => string.Join("", a.ProteinInterfaceAminoAcids1L().Select((b, i) => !a.InteractionBools()[i] ? b : ' ').ToList())).ToList();
var distinctProteinInterfaceLengths = vectorProteinInterfaceWholeList.Select(a => a.ProteinInterfaceLength).Distinct().ToArray();
var directionFwd = vectorProteinInterfaceWholeList.Count(a => !a.ReversedSequence);
var directionRev = vectorProteinInterfaceWholeList.Count - directionFwd;
string direction = "";
if (directionFwd > 0 && directionRev == 0)
{
direction = "Fwd";
}
else if (directionFwd == 0 && directionRev > 0)
{
direction = "Rev";
}
else
{
direction = "Mix";
}
var record = new ProproteinInterfaceSpreadsheetRecord
{
MotifName = "",
MotifSource = "",
TotalFound = "" + vectorProteinInterfaceWholeList.Count,
Direction = direction,
ProteinInterfaceLength = "" + string.Join(", ", distinctProteinInterfaceLengths),
MotifAminoAcids = ProproteinInterfaceMotif.FindProteinInterfaceAminoAcidMotif(proteinInterfaceSequenceList),
MotifInteractionAminoAcids = ProproteinInterfaceMotif.FindProteinInterfaceAminoAcidMotif(proteinInterfaceInteractionSequenceList),
MotifNonInteractionAminoAcids = ProproteinInterfaceMotif.FindProteinInterfaceAminoAcidMotif(proteinInterfaceNonInteractionSequenceList),
MotifPhysicochemical = ProproteinInterfaceMotif.FindProteinInterfaceAminoAcidGroupMotif(AminoAcidGroups.AminoAcidGroups.EnumAminoAcidGroups.Physicochemical, proteinInterfaceSequenceList),
MotifInteractionPhysicochemical = ProproteinInterfaceMotif.FindProteinInterfaceAminoAcidGroupMotif(AminoAcidGroups.AminoAcidGroups.EnumAminoAcidGroups.Physicochemical, proteinInterfaceInteractionSequenceList),
MotifNonInteractionPhysicochemical = ProproteinInterfaceMotif.FindProteinInterfaceAminoAcidGroupMotif(AminoAcidGroups.AminoAcidGroups.EnumAminoAcidGroups.Physicochemical, proteinInterfaceNonInteractionSequenceList),
MotifHydrophobicity = ProproteinInterfaceMotif.FindProteinInterfaceAminoAcidGroupMotif(AminoAcidGroups.AminoAcidGroups.EnumAminoAcidGroups.Hydrophobicity, proteinInterfaceSequenceList),
MotifInteractionHydrophobicity = ProproteinInterfaceMotif.FindProteinInterfaceAminoAcidGroupMotif(AminoAcidGroups.AminoAcidGroups.EnumAminoAcidGroups.Hydrophobicity, proteinInterfaceInteractionSequenceList),
MotifNonInteractionHydrophobicity = ProproteinInterfaceMotif.FindProteinInterfaceAminoAcidGroupMotif(AminoAcidGroups.AminoAcidGroups.EnumAminoAcidGroups.Hydrophobicity, proteinInterfaceNonInteractionSequenceList),
MotifPdbSum = ProproteinInterfaceMotif.FindProteinInterfaceAminoAcidGroupMotif(AminoAcidGroups.AminoAcidGroups.EnumAminoAcidGroups.PdbSum, proteinInterfaceSequenceList),
MotifInteractionPdbSum = ProproteinInterfaceMotif.FindProteinInterfaceAminoAcidGroupMotif(AminoAcidGroups.AminoAcidGroups.EnumAminoAcidGroups.PdbSum, proteinInterfaceInteractionSequenceList),
MotifNonInteractionPdbSum = ProproteinInterfaceMotif.FindProteinInterfaceAminoAcidGroupMotif(AminoAcidGroups.AminoAcidGroups.EnumAminoAcidGroups.PdbSum, proteinInterfaceNonInteractionSequenceList),
MotifUniProtKb = ProproteinInterfaceMotif.FindProteinInterfaceAminoAcidGroupMotif(AminoAcidGroups.AminoAcidGroups.EnumAminoAcidGroups.UniProtKb, proteinInterfaceSequenceList),
MotifInteractionUniProtKb = ProproteinInterfaceMotif.FindProteinInterfaceAminoAcidGroupMotif(AminoAcidGroups.AminoAcidGroups.EnumAminoAcidGroups.UniProtKb, proteinInterfaceInteractionSequenceList),
MotifNonInteractionUniProtKb = ProproteinInterfaceMotif.FindProteinInterfaceAminoAcidGroupMotif(AminoAcidGroups.AminoAcidGroups.EnumAminoAcidGroups.UniProtKb, proteinInterfaceNonInteractionSequenceList),
};
for (var index = record.MotifCommonProperties.Length - 1; index >= 6; index--)
{
record.MotifCommonProperties[index] = ProproteinInterfaceMotif.FindProteinInterfaceAminoAcidCommonPropertiesMotif(proteinInterfaceSequenceList, AminoAcidPropertyMatchType.MininumMatch, index + 1);
record.MotifInteractionsCommonProperties[index] = ProproteinInterfaceMotif.FindProteinInterfaceAminoAcidCommonPropertiesMotif(proteinInterfaceInteractionSequenceList, AminoAcidPropertyMatchType.MininumMatch, index + 1);
record.MotifNonInteractionsCommonProperties[index] = ProproteinInterfaceMotif.FindProteinInterfaceAminoAcidCommonPropertiesMotif(proteinInterfaceNonInteractionSequenceList, AminoAcidPropertyMatchType.MininumMatch, index + 1);
}
return(record);
}
