/// <summary>
/// This method returns a list of amino acid codes that match any of the specified true properties. If all parameter properties are false, it returns a list of
/// </summary>
/// <param name="aminoAcidData"></param>
/// <param name="aminoAcidPropertyMatchType"></param>
/// <param name="matchValue"></param>
/// <returns></returns>
public static string ListAminoAcidsByProperty(AminoAcidProperties <bool> aminoAcidData, AminoAcidPropertyMatchType aminoAcidPropertyMatchType = AminoAcidPropertyMatchType.AnyTrueMatch, double matchValue = 0)
{
var result = new List <string>();
var aminoAcidDataPropertiesArray = aminoAcidData.PropertiesArray();
for (var index = 0; index < AminoAcidTotals.TotalAminoAcids(); index++)
{
var aa = AminoAcidConversions.AminoAcidNumberToAminoAcidObject(index + 1);
//aminoAcidDataPropertiesArray = aa.PropertiesArray();
var propertyMatches = new AminoAcidProperties <bool>()
{
//Acidic = aminoAcidData.Acidic == aa.Acidic,
Aliphatic = aminoAcidData.Aliphatic == aa.Aliphatic,
Aromatic = aminoAcidData.Aromatic == aa.Aromatic,
Charged = aminoAcidData.Charged == aa.Charged,
Hydrophobic = aminoAcidData.Hydrophobic == aa.Hydrophobic,
Hydroxylic = aminoAcidData.Hydroxylic == aa.Hydroxylic,
Negative = aminoAcidData.Negative == aa.Negative,
Polar = aminoAcidData.Polar == aa.Polar,
Positive = aminoAcidData.Positive == aa.Positive,
Small = aminoAcidData.Small == aa.Small,
Sulphur = aminoAcidData.Sulphur == aa.Sulphur,
Tiny = aminoAcidData.Tiny == aa.Tiny,
};
var matchedPropertiesArray = propertyMatches.PropertiesArray();
// number of equal boolean values in aminoAcidData and aa
var totalMatchingProperties = matchedPropertiesArray.Where(a => a).Count();
var totalTrueMatchingProperties = aminoAcidDataPropertiesArray.Where((a, i) => a && matchedPropertiesArray[i]).Count();
var totalFalseMatchingProperties = aminoAcidDataPropertiesArray.Where((a, i) => !a && matchedPropertiesArray[i]).Count();
var totalTrueParameterProperties = aminoAcidDataPropertiesArray.Count(a => a);
var totalFalseParameterProperties = aminoAcidDataPropertiesArray.Count(a => !a);
var percentageMatchingProperties = totalMatchingProperties / matchedPropertiesArray.Length;
var percentageTrueMatchingProperties = totalTrueParameterProperties > 0 ? totalTrueMatchingProperties / totalTrueParameterProperties : 0;
var percentageFalseMatchingProperties = totalFalseParameterProperties > 0 ? totalFalseMatchingProperties / totalFalseParameterProperties : 0;
var noPropertyMatches = totalMatchingProperties == 0;
var anyPropertyMatches = totalMatchingProperties > 0;
var allPropertiesMatch = matchedPropertiesArray.Length == totalMatchingProperties;
var anyTrueMatch = aminoAcidDataPropertiesArray.Where((a, i) => a && matchedPropertiesArray[i]).Any();
var anyFalseMatch = aminoAcidDataPropertiesArray.Where((a, i) => !a && matchedPropertiesArray[i]).Any();
var allTrueMatch = aminoAcidDataPropertiesArray.Where((a, i) => a && matchedPropertiesArray[i]).Count() == aminoAcidDataPropertiesArray.Length;
var allFalseMatch = aminoAcidDataPropertiesArray.Where((a, i) => !a && matchedPropertiesArray[i]).Count() == aminoAcidDataPropertiesArray.Length;
if ((aminoAcidPropertyMatchType == AminoAcidPropertyMatchType.AllMatch && allPropertiesMatch) ||
(aminoAcidPropertyMatchType == AminoAcidPropertyMatchType.AllTrueMatch && allTrueMatch) ||
(aminoAcidPropertyMatchType == AminoAcidPropertyMatchType.AllFalseMatch && allFalseMatch) ||
(aminoAcidPropertyMatchType == AminoAcidPropertyMatchType.AnyMatch && anyPropertyMatches) ||
(aminoAcidPropertyMatchType == AminoAcidPropertyMatchType.AnyTrueMatch && anyTrueMatch) ||
(aminoAcidPropertyMatchType == AminoAcidPropertyMatchType.AnyFalseMatch && anyFalseMatch) ||
(aminoAcidPropertyMatchType == AminoAcidPropertyMatchType.NoneMatch && noPropertyMatches) ||
(aminoAcidPropertyMatchType == AminoAcidPropertyMatchType.MininumMatch && totalMatchingProperties >= matchValue) ||
(aminoAcidPropertyMatchType == AminoAcidPropertyMatchType.MininumTrueMatch && totalTrueMatchingProperties >= matchValue) ||
(aminoAcidPropertyMatchType == AminoAcidPropertyMatchType.MininumFalseMatch && totalFalseMatchingProperties >= matchValue) ||
(aminoAcidPropertyMatchType == AminoAcidPropertyMatchType.PercentageMatch && percentageMatchingProperties >= matchValue) ||
(aminoAcidPropertyMatchType == AminoAcidPropertyMatchType.PercentageTrueMatch && percentageTrueMatchingProperties >= matchValue) ||
(aminoAcidPropertyMatchType == AminoAcidPropertyMatchType.PercentageFalseMatch && percentageFalseMatchingProperties >= matchValue))
{
result.Add(aa.Code1L);
}
}
var resultStr = string.Join("", result.OrderBy(a => a).ToArray());
return(resultStr);
}
public static string FindProteinInterfaceAminoAcidCommonPropertiesMotif(List <string> proteinInterfaceSequenceList, AminoAcidPropertyMatchType aminoAcidPropertyMatchType, double matchValue = 0)
{
if (proteinInterfaceSequenceList == null)
{
throw new ArgumentNullException(nameof(proteinInterfaceSequenceList));
}
var totalProteinInterfaces = proteinInterfaceSequenceList.Count;
var proteinInterfaceLength = proteinInterfaceSequenceList.Select(a => a.Length).Max();
var aminoAcidObjects = new AminoAcidProperties <bool> [TotalAminoAcids];
for (var aaIndex = 0; aaIndex < aminoAcidObjects.Length; aaIndex++)
{
aminoAcidObjects[aaIndex] = AminoAcidConversions.AminoAcidNumberToAminoAcidObject(aaIndex + 1);
}
var aminoAcidPositionCount = new AminoAcidProperties <int> [proteinInterfaceLength];
for (var position = 0; position < aminoAcidPositionCount.Length; position++)
{
aminoAcidPositionCount[position] = new AminoAcidProperties <int>();
}
var motifs = new List <string>();
for (int positionIndex = 0; positionIndex < proteinInterfaceLength; positionIndex++)
{
string positionMotif = "";
foreach (var aaCode in proteinInterfaceSequenceList.Where(a => a.Length > positionIndex).Select(a => a[positionIndex]).Distinct().ToArray())
{
if (!char.IsLetterOrDigit(aaCode) || NonStandardAminoAcids.Contains(aaCode))
{
continue;
}
var aminoAcid = aminoAcidObjects[AminoAcidConversions.AminoAcidNameToNumber(aaCode) - 1];
var matches = AminoAcidConversions.ListAminoAcidsByProperty(aminoAcid, aminoAcidPropertyMatchType, matchValue);
positionMotif += string.Join("", matches.Where(a => !positionMotif.Contains(a) && !NonStandardAminoAcids.Contains(a)).ToArray());
}
motifs.Add(positionMotif);
}
return(MotifFormatter(motifs));
}