public static void FillInAnnotation(IDictionary <string, ProteinAnnotation> annotations,
IDictionary <string, MiniProteinAnnotation> map, IDictionary <string, string> pspMap)
{
foreach (ProteinAnnotation pa in annotations.Values)
{
if (map.ContainsKey(pa.Accession))
{
MiniProteinAnnotation mpa = map[pa.Accession];
pa.GeneName = mpa.GeneName;
pa.IsReviewed = mpa.IsReviewed;
pa.ProteinName = mpa.ProteinName;
pa.PfamIds = mpa.PfamIds;
pa.PfamNames = mpa.PfamNames;
pa.PfamStart = mpa.PfamStart;
pa.PfamEnd = mpa.PfamEnd;
pa.Pdbs = mpa.Pdbs;
pa.Features = mpa.Features;
}
else if (pa.Accession.Contains("-"))
{
string acc = pa.Accession.Substring(0, pa.Accession.IndexOf('-'));
if (map.ContainsKey(acc))
{
MiniProteinAnnotation mpa = map[acc];
pa.GeneName = mpa.GeneName;
pa.IsReviewed = mpa.IsReviewed;
pa.ProteinName = mpa.ProteinName;
pa.PfamIds = mpa.PfamIds;
pa.PfamNames = mpa.PfamNames;
pa.PfamStart = new int[0];
pa.PfamEnd = new int[0];
pa.Pdbs = new string[0];
pa.Features = new Dictionary <FeatureType, List <UniprotFeature> >();
}
}
if (pspMap.ContainsKey(pa.Accession))
{
string x = pspMap[pa.Accession];
string[] modificationsPsp = x.Length == 0 ? new string[0] : x.Split(';');
string[] modTypesPsp = new string[modificationsPsp.Length];
int[] modPosPsp = new int[modificationsPsp.Length];
for (int i = 0; i < modificationsPsp.Length; i++)
{
string[] w = modificationsPsp[i].Split(',');
modTypesPsp[i] = w[0];
bool success = Parser.TryInt(w[1], out modPosPsp[i]);
if (!success)
{
modPosPsp[i] = -1;
}
}
pa.ModTypesPsp = modTypesPsp;
pa.ModPosPsp = modPosPsp;
}
}
}
public static void FillInAnnotation(IDictionary <string, ProteinAnnotation> annots)
{
string file1 = Path.Combine(FileUtils.GetConfigPath(), "maxquantAnnot.txt");
string file2 = file1 + ".gz";
string file = null;
if (File.Exists(file1))
{
file = file1;
}
else if (File.Exists(file2))
{
file = file2;
}
if (file != null)
{
HashSet <string> acc = GetAccessions(annots);
Dictionary <string, MiniProteinAnnotation> map = MiniProteinAnnotation.ReadMapping(file, acc);
string phosphoSiteFolder = FileUtils.executablePath + "\\conf\\PSP";
Dictionary <string, string> psMap = PhosphoSitePlusParser.ParsePhosphoSite(phosphoSiteFolder);
FillInAnnotation(annots, map, psMap);
}
}
public void ProcessData(IMatrixData mdata, Parameters param, ref IMatrixData[] supplTables,
ref IDocumentData[] documents, ProcessInfo processInfo)
{
string folder = FileUtils.executablePath + "\\conf";
string file = folder + "\\maxquantAnnot.txt.gz";
int protInd = param.GetParam <int>("Proteins").Value;
int posInd = param.GetParam <int>("Positions within proteins").Value;
bool addStatus = param.GetParam <bool>("Add status column").Value;
string[] protCol = mdata.StringColumns[protInd];
HashSet <string> allProtIds = new HashSet <string>();
string[][] protIds = new string[protCol.Length][];
for (int i = 0; i < protCol.Length; i++)
{
protIds[i] = protCol[i].Length > 0 ? protCol[i].Split(';') : new string[0];
foreach (string s in protIds[i])
{
if (!allProtIds.Contains(s))
{
allProtIds.Add(s);
}
}
}
Dictionary <string, MiniProteinAnnotation> map = MiniProteinAnnotation.ReadMapping(file, allProtIds);
string[] posCol = mdata.StringColumns[posInd];
int nrows = protCol.Length;
string[][] pfamCol = new string[nrows][];
Dictionary <FeatureType, string[][]> cols = new Dictionary <FeatureType, string[][]>();
Dictionary <FeatureType, string[][]> statusCols = new Dictionary <FeatureType, string[][]>();
foreach (FeatureType t in FeatureType.allFeatureTypes)
{
cols.Add(t, new string[nrows][]);
statusCols.Add(t, new string[nrows][]);
}
for (int i = 0; i < protCol.Length; i++)
{
string[] posString = posCol[i].Length > 0 ? posCol[i].Split(';') : new string[0];
HashSet <string> pfams = new HashSet <string>();
Dictionary <FeatureType, HashSet <string> > others = new Dictionary <FeatureType, HashSet <string> >();
Dictionary <FeatureType, HashSet <string> > othersStatus = new Dictionary <FeatureType, HashSet <string> >();
for (int j = 0; j < protIds[i].Length; j++)
{
string protId = protIds[i][j];
int pos = Parser.Int(posString[j]);
if (map.ContainsKey(protId))
{
MiniProteinAnnotation mpa = map[protId];
for (int k = 0; k < mpa.PfamIds.Length; k++)
{
if (Fits(pos, mpa.PfamStart[k], mpa.PfamEnd[k]))
{
pfams.Add(mpa.PfamNames[k]);
}
}
foreach (FeatureType featureType in mpa.Features.Keys)
{
foreach (UniprotFeature uf in mpa.Features[featureType])
{
if (!Parser.TryInt(uf.FeatureBegin, out int begin))
{
begin = int.MaxValue;
}
if (!Parser.TryInt(uf.FeatureEnd, out int end))
{
end = int.MinValue;
}
if (Fits(pos, begin, end))
{
if (!others.ContainsKey(featureType))
{
others.Add(featureType, new HashSet <string>());
othersStatus.Add(featureType, new HashSet <string>());
}
string x = uf.FeatureDescription;
if (string.IsNullOrEmpty(x))
{
x = "+";
}
others[featureType].Add(x);
string y = uf.FeatureStatus;
if (!string.IsNullOrEmpty(y))
{
othersStatus[featureType].Add(y);
}
}
}
}
}
}
pfamCol[i] = ToArray(pfams);
foreach (FeatureType t in FeatureType.allFeatureTypes)
{
if (others.ContainsKey(t))
{
cols[t][i] = ToArray(others[t]);
}
else
{
cols[t][i] = new string[0];
}
if (othersStatus.ContainsKey(t))
{
statusCols[t][i] = ToArray(othersStatus[t]);
}
else
{
statusCols[t][i] = new string[0];
}
}
}
mdata.AddCategoryColumn("Pfam domains", "", pfamCol);
foreach (FeatureType t in FeatureType.allFeatureTypes)
{
mdata.AddCategoryColumn(t.UniprotName, "", cols[t]);
if (addStatus)
{
mdata.AddCategoryColumn(t.UniprotName + " status", "", statusCols[t]);
}
}
}