public Parameters GetParameters(IMatrixData mdata, ref string errorString)
{
List <string> colChoice = mdata.StringColumnNames;
int colInd = 0;
for (int i = 0; i < colChoice.Count; i++)
{
if (colChoice[i].ToUpper().Equals("UNIPROT"))
{
colInd = i;
break;
}
}
string[] choice = PhosphoSitePlusParser.GetAllMods();
return
(new Parameters(
new MultiChoiceParam("Modifications")
{
Value = ArrayUtils.ConsecutiveInts(choice.Length), Values = choice
},
new SingleChoiceParam("Uniprot column")
{
Value = colInd,
Help = "Specify here the column that contains Uniprot identifiers.",
Values = colChoice
}));
}
public Parameters GetParameters(IMatrixData mdata, ref string errorString)
{
List <string> colChoice = mdata.StringColumnNames;
int colInd = 0;
for (int i = 0; i < colChoice.Count; i++)
{
if (colChoice[i].ToUpper().Equals("UNIPROT"))
{
colInd = i;
break;
}
}
int colSeqInd = 0;
for (int i = 0; i < colChoice.Count; i++)
{
if (colChoice[i].ToUpper().Equals("SEQUENCE WINDOW"))
{
colSeqInd = i;
break;
}
}
string[] choice = PhosphoSitePlusParser.GetAllMods();
int ind = ArrayUtils.IndexOf(choice, "Phosphorylation");
return
(new Parameters(
new SingleChoiceParam("Modification")
{
Value = ind,
Values = choice,
Help = "Select here the kind of modification for which information should be added."
},
new SingleChoiceParam("Uniprot column")
{
Value = colInd,
Help = "Specify here the column that contains Uniprot identifiers.",
Values = colChoice
},
new SingleChoiceParam("Sequence window")
{
Value = colSeqInd,
Help = "Specify here the column that contains the sequence windows around the site.",
Values = colChoice
}));
}
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 mod = param.GetParam <int>("Modification").StringValue;
PhosphoSitePlusParser.ParseKnownMod(mod, out string[] seqWins, out string[] accs, out string[] pubmedLtp, out string[] pubmedMs2, out string[] cstMs2, out string[] species);
if (seqWins == null)
{
processInfo.ErrString = "File does not exist.";
return;
}
string[] up = mdata.StringColumns[param.GetParam <int>("Uniprot column").Value];
string[][] uprot = new string[up.Length][];
for (int i = 0; i < up.Length; i++)
{
uprot[i] = up[i].Length > 0 ? up[i].Split(';') : new string[0];
}
string[] win = mdata.StringColumns[param.GetParam <int>("Sequence window").Value];
Dictionary <string, List <int> > map = new Dictionary <string, List <int> >();
for (int i = 0; i < seqWins.Length; i++)
{
string acc = accs[i];
if (!map.ContainsKey(acc))
{
map.Add(acc, new List <int>());
}
map[acc].Add(i);
}
string[] newCol = new string[uprot.Length];
string[][] newCatCol = new string[uprot.Length][];
string[][] originCol = new string[uprot.Length][];
for (int i = 0; i < newCol.Length; i++)
{
string[] win1 = TransformIl(win[i]).Split(';');
HashSet <string> wins = new HashSet <string>();
HashSet <string> origins = new HashSet <string>();
foreach (string ux in uprot[i])
{
if (map.ContainsKey(ux))
{
List <int> n = map[ux];
foreach (int ind in n)
{
string s = seqWins[ind];
if (Contains(win1, TransformIl(s.ToUpper().Substring(1, s.Length - 2))))
{
wins.Add(s);
if (pubmedLtp[ind].Length > 0)
{
origins.Add("LTP");
}
if (pubmedMs2[ind].Length > 0)
{
origins.Add("HTP");
}
if (cstMs2[ind].Length > 0)
{
origins.Add("CST");
}
}
}
}
}
if (wins.Count > 0)
{
newCol[i] = StringUtils.Concat(";", ArrayUtils.ToArray(wins));
newCatCol[i] = new[] { "+" };
string[] x = ArrayUtils.ToArray(origins);
Array.Sort(x);
originCol[i] = x;
}
else
{
newCol[i] = "";
newCatCol[i] = new string[0];
originCol[i] = new string[0];
}
}
mdata.AddStringColumn("PhosphoSitePlus window", "", newCol);
mdata.AddCategoryColumn("Known site", "", newCatCol);
mdata.AddCategoryColumn("Origin", "", originCol);
}
public void ProcessData(IMatrixData mdata, Parameters param, ref IMatrixData[] supplTables,
ref IDocumentData[] documents, ProcessInfo processInfo)
{
string[] mods = param.GetParam <int[]>("Modifications").StringValue.Split(new[] { ';' },
StringSplitOptions.RemoveEmptyEntries);
string[] up = mdata.StringColumns[param.GetParam <int>("Uniprot column").Value];
string[][] uprot = new string[up.Length][];
for (int i = 0; i < up.Length; i++)
{
uprot[i] = up[i].Length > 0 ? up[i].Split(';') : new string[0];
}
double[][] c = new double[mods.Length][];
for (int index = 0; index < mods.Length; index++)
{
string mod = mods[index];
string filename = PhosphoSitePlusParser.GetFilenameForMod(mod);
if (filename == null)
{
processInfo.ErrString = "File does not exist.";
return;
}
PhosphoSitePlusParser.ParseKnownMods(filename, out string[] seqWins, out string[] accs, out string[] pubmedLtp, out string[] pubmedMs2, out string[] cstMs2, out string[] species);
for (int i = 0; i < seqWins.Length; i++)
{
seqWins[i] = seqWins[i].ToUpper();
}
Dictionary <string, HashSet <string> > counts = new Dictionary <string, HashSet <string> >();
for (int i = 0; i < accs.Length; i++)
{
string acc = accs[i];
if (!counts.ContainsKey(acc))
{
counts.Add(acc, new HashSet <string>());
}
counts[acc].Add(seqWins[i]);
}
c[index] = new double[up.Length];
for (int i = 0; i < up.Length; i++)
{
c[index][i] = CountSites(uprot[i], counts);
}
}
string[][] catCol = new string[up.Length][];
for (int i = 0; i < catCol.Length; i++)
{
List <string> x = new List <string>();
for (int j = 0; j < mods.Length; j++)
{
if (c[j][i] > 0)
{
x.Add(mods[j]);
}
}
x.Sort();
catCol[i] = x.ToArray();
}
mdata.AddCategoryColumn("Known modifications", "Known modifications", catCol);
for (int i = 0; i < mods.Length; i++)
{
mdata.AddNumericColumn(mods[i] + " count", mods[i] + " count", c[i]);
}
}
public void ProcessData(IMatrixData mdata, Parameters param, ref IMatrixData[] supplTables,
ref IDocumentData[] documents, ProcessInfo processInfo)
{
string[] seqWins;
string[] accs;
string[] function;
string[] process;
string[] protInteract;
string[] otherInteract;
string[] notes;
string[] species;
PhosphoSitePlusParser.ParseRegulatorySites(out seqWins, out accs, out function, out process, out protInteract,
out otherInteract, out notes, out species);
if (seqWins == null)
{
processInfo.ErrString = "File does not exist.";
return;
}
string[] up = mdata.StringColumns[param.GetParam <int>("Uniprot column").Value];
string[][] uprot = new string[up.Length][];
for (int i = 0; i < up.Length; i++)
{
uprot[i] = up[i].Length > 0 ? up[i].Split(';') : new string[0];
}
string[] win = mdata.StringColumns[param.GetParam <int>("Sequence window").Value];
Dictionary <string, List <int> > map = new Dictionary <string, List <int> >();
for (int i = 0; i < seqWins.Length; i++)
{
string acc = accs[i];
if (!map.ContainsKey(acc))
{
map.Add(acc, new List <int>());
}
map[acc].Add(i);
}
string[][] newCatCol = new string[uprot.Length][];
string[][] function2 = new string[uprot.Length][];
string[][] process2 = new string[uprot.Length][];
string[][] protInteract2 = new string[uprot.Length][];
string[][] otherInteract2 = new string[uprot.Length][];
string[][] notes2 = new string[uprot.Length][];
for (int i = 0; i < uprot.Length; i++)
{
string[] win1 = TransformIl(win[i]).Split(';');
HashSet <string> wins = new HashSet <string>();
HashSet <string> function1 = new HashSet <string>();
HashSet <string> process1 = new HashSet <string>();
HashSet <string> protInteract1 = new HashSet <string>();
HashSet <string> otherInteract1 = new HashSet <string>();
HashSet <string> notes1 = new HashSet <string>();
foreach (string ux in uprot[i])
{
if (map.ContainsKey(ux))
{
List <int> n = map[ux];
foreach (int ind in n)
{
string s = seqWins[ind];
if (Contains(win1, TransformIl(s.ToUpper().Substring(1, s.Length - 2))))
{
wins.Add(s);
if (function[ind].Length > 0)
{
function1.Add(function[ind]);
}
if (process[ind].Length > 0)
{
process1.Add(process[ind]);
}
if (protInteract[ind].Length > 0)
{
protInteract1.Add(protInteract[ind]);
}
if (otherInteract[ind].Length > 0)
{
otherInteract1.Add(otherInteract[ind]);
}
if (notes[ind].Length > 0)
{
notes1.Add(notes[ind]);
}
}
}
}
}
if (wins.Count > 0)
{
newCatCol[i] = new[] { "+" };
function2[i] = ArrayUtils.ToArray(function1);
process2[i] = ArrayUtils.ToArray(process1);
protInteract2[i] = ArrayUtils.ToArray(protInteract1);
otherInteract2[i] = ArrayUtils.ToArray(otherInteract1);
notes2[i] = ArrayUtils.ToArray(notes1);
}
else
{
newCatCol[i] = new string[0];
function2[i] = new string[0];
process2[i] = new string[0];
protInteract2[i] = new string[0];
otherInteract2[i] = new string[0];
notes2[i] = new string[0];
}
}
mdata.AddCategoryColumn("Regulatory site", "", newCatCol);
mdata.AddCategoryColumn("Regulatory site function", "", function2);
mdata.AddCategoryColumn("Regulatory site process", "", process2);
mdata.AddCategoryColumn("Regulatory site protInteract", "", protInteract2);
mdata.AddCategoryColumn("Regulatory site otherInteract", "", otherInteract2);
mdata.AddCategoryColumn("Regulatory site notes", "", notes2);
}
public void ProcessData(IMatrixData mdata, Parameters param, ref IMatrixData[] supplTables,
ref IDocumentData[] documents, ProcessInfo processInfo)
{
string[] seqWins;
string[] subAccs;
string[] kinases;
string[] kinAccs;
string[] species;
PhosphoSitePlusParser.ParseKinaseSubstrate(out seqWins, out subAccs, out kinases, out kinAccs, out species);
if (seqWins == null)
{
processInfo.ErrString = "File does not exist.";
return;
}
string[] up = mdata.StringColumns[param.GetParam <int>("Uniprot column").Value];
string[][] uprot = new string[up.Length][];
for (int i = 0; i < up.Length; i++)
{
uprot[i] = up[i].Length > 0 ? up[i].Split(';') : new string[0];
}
string[] win = mdata.StringColumns[param.GetParam <int>("Sequence window").Value];
Dictionary <string, List <Tuple <string, string, string> > > substrateProperties =
new Dictionary <string, List <Tuple <string, string, string> > >();
for (int i = 0; i < seqWins.Length; i++)
{
string subAcc = subAccs[i];
if (!substrateProperties.ContainsKey(subAcc))
{
substrateProperties.Add(subAcc, new List <Tuple <string, string, string> >());
}
substrateProperties[subAcc].Add(new Tuple <string, string, string>(seqWins[i], kinases[i], kinAccs[i]));
}
string[] kinaseNameColumn = new string[uprot.Length];
string[] kinaseUniprotColumn = new string[uprot.Length];
for (int i = 0; i < kinaseNameColumn.Length; i++)
{
string[] win1 = AddKnownSites.TransformIl(win[i]).Split(';');
HashSet <string> kinaseNamesHits = new HashSet <string>();
HashSet <string> kinaseUniprotHits = new HashSet <string>();
foreach (string ux in uprot[i])
{
if (substrateProperties.ContainsKey(ux))
{
List <Tuple <string, string, string> > properties = substrateProperties[ux];
foreach (Tuple <string, string, string> property in properties)
{
string w = property.Item1;
if (AddKnownSites.Contains(win1, AddKnownSites.TransformIl(w.ToUpper().Substring(1, w.Length - 2))))
{
kinaseNamesHits.Add(property.Item2);
kinaseUniprotHits.Add(property.Item3);
}
}
}
}
kinaseNameColumn[i] = kinaseNamesHits.Count > 0 ? StringUtils.Concat(";", ArrayUtils.ToArray(kinaseNamesHits)) : "";
kinaseUniprotColumn[i] = kinaseUniprotHits.Count > 0
? StringUtils.Concat(";", ArrayUtils.ToArray(kinaseUniprotHits))
: "";
}
mdata.AddStringColumn("PhosphoSitePlus kinase", "", kinaseNameColumn);
mdata.AddStringColumn("PhosphoSitePlus kinase uniprot", "", kinaseUniprotColumn);
}