public void ReadMappingTest()
{
Assert.Inconclusive("Should be moved to integration tests, using conf");
//[DeploymentItem("conf", "conf")]
string[] baseNames, files;
var annots = PerseusUtils.GetAvailableAnnots(out baseNames, out files);
var uniprotIndex = baseNames.ToList().FindIndex(name => name.ToLower().Equals("uniprot"));
var selection = annots[uniprotIndex].ToList().FindIndex(name => name.ToLower().Equals("gene name"));
var ids = new[] { "P08908" };
var mapping = AddAnnotationToMatrix.ReadMapping(ids, files[1], new[] { selection });
CollectionAssert.AreEqual(new [] { "HTR1A" }, mapping[ids[0]]);
}
private static string[] GetBaseIds(Parameters para, IDataWithAnnotationColumns mdata)
{
string[] baseNames;
AnnotType[][] types;
string[] files;
PerseusUtils.GetAvailableAnnots(out baseNames, out types, out files);
ParameterWithSubParams <int> spd = para.GetParamWithSubParams <int>("Source");
int ind = spd.Value;
Parameters param = spd.GetSubParameters();
int baseCol = param.GetParam <int>(baseNames[ind] + " column").Value;
string[] baseIds = mdata.StringColumns[baseCol];
return(baseIds);
}
public void GetAvailableAnnotsTest()
{
string[] annotFiles = PerseusUtils.GetAnnotFiles();
Assert.Inconclusive("Should be moved to integration tests, using conf");
//[DeploymentItem("conf", "conf")]
string[] baseNames;
string[] files;
string[][] annots = PerseusUtils.GetAvailableAnnots(out baseNames, out files);
Assert.AreEqual(3, files.Length);
Assert.AreEqual(3, baseNames.Length);
Assert.AreEqual(3, annots.Length);
CollectionAssert.AreEqual(new[] { "ENSG", "UniProt", "ENSG" }, baseNames);
CollectionAssert.AreEqual(new[] { "Chromosome", "Base pair index", "Orientation" }, annots[0]);
}
private static string[][] GetAvailableTextAnnots(out string[] baseNames, out int[][] inds, out string[] files)
{
AnnotType[][] types;
string[][] annots = PerseusUtils.GetAvailableAnnots(out baseNames, out types, out files);
inds = new int[files.Length][];
for (int i = 0; i < files.Length; i++)
{
List <int> result = new List <int>();
for (int j = 0; j < types[i].Length; j++)
{
if (types[i][j] == AnnotType.Text)
{
result.Add(j);
}
}
inds[i] = result.ToArray();
annots[i] = ArrayUtils.SubArray(annots[i], result);
}
return(annots);
}
public Parameters GetParameters(IMatrixData mdata, ref string errorString)
{
List <string> colChoice = mdata.StringColumnNames;
string[] baseNames;
string[] files;
List <string> badFiles;
string[][] annots = PerseusUtils.GetAvailableAnnots(out baseNames, out files, out badFiles);
if (badFiles.Any())
{
errorString = $"Could not load annotations from file(s): {string.Join(", ", badFiles)}";
}
int selFile = 0;
bool isMainAnnot = false;
for (int i = 0; i < files.Length; i++)
{
if (files[i].ToLower().Contains("perseusannot"))
{
selFile = i;
isMainAnnot = true;
break;
}
}
Parameters[] subParams = new Parameters[files.Length];
for (int i = 0; i < subParams.Length; i++)
{
int colInd = 0;
if (isMainAnnot && i == selFile)
{
for (int j = 0; j < colChoice.Count; j++)
{
if (colChoice[j].ToUpper().Contains("PROTEIN IDS"))
{
colInd = j;
break;
}
}
for (int j = 0; j < colChoice.Count; j++)
{
if (colChoice[j].ToUpper().Contains("MAJORITY PROTEIN IDS"))
{
colInd = j;
break;
}
}
}
else
{
for (int j = 0; j < colChoice.Count; j++)
{
if (colChoice[j].ToUpper().Contains(baseNames[i].ToUpper()))
{
colInd = j;
break;
}
}
}
subParams[i] =
new Parameters(
new SingleChoiceParam(baseNames[i] + " column")
{
Values = colChoice,
Value = colInd,
Help =
"Specify here the column that contains the base identifiers which are going to be " +
"matched to the annotation."
}, new MultiChoiceParam("Annotations to be added")
{
Values = annots[i]
});
}
return
(new Parameters(
new SingleChoiceWithSubParams("Source", selFile)
{
Values = files,
SubParams = subParams,
ParamNameWidth = 136,
TotalWidth = 735
}, new MultiChoiceParam("Additional sources")
{
Values = files
}));
}
public static bool ProcessDataAddAnnotation(int nrows, Parameters para, string[] baseIds, ProcessInfo processInfo,
out string[] name, out int[] catColInds, out int[] textColInds, out int[] numColInds, out string[][][] catCols,
out string[][] textCols, out double[][] numCols)
{
string[] baseNames;
AnnotType[][] types;
string[] files;
string[][] names = PerseusUtils.GetAvailableAnnots(out baseNames, out types, out files);
const bool deHyphenate = true;
ParameterWithSubParams <int> spd = para.GetParamWithSubParams <int>("Source");
int ind = spd.Value;
Parameters param = spd.GetSubParameters();
AnnotType[] type = types[ind];
name = names[ind];
int[] addtlSources = para.GetParam <int[]>("Additional sources").Value;
addtlSources = ArrayUtils.Remove(addtlSources, ind);
foreach (int addtlSource in addtlSources)
{
AnnotType[] type1 = types[addtlSource];
string[] name1 = names[addtlSource];
if (!ArrayUtils.EqualArrays(type, type1))
{
processInfo.ErrString = "Additional annotation file does not have the same column structure.";
catColInds = new int[] {};
textColInds = new int[] {};
numColInds = new int[] {};
catCols = new string[][][] {};
textCols = new string[][] {};
numCols = new double[][] {};
return(false);
}
if (!ArrayUtils.EqualArrays(name, name1))
{
processInfo.ErrString = "Additional annotation file does not have the same column structure.";
catColInds = new int[] {};
textColInds = new int[] {};
numColInds = new int[] {};
catCols = new string[][][] {};
textCols = new string[][] {};
numCols = new double[][] {};
return(false);
}
}
int[] selection = param.GetParam <int[]>("Annotations to be added").Value;
type = ArrayUtils.SubArray(type, selection);
name = ArrayUtils.SubArray(name, selection);
HashSet <string> allIds = GetAllIds(baseIds, deHyphenate);
Dictionary <string, string[]> mapping = ReadMapping(allIds, files[ind], selection);
foreach (int addtlSource in addtlSources)
{
Dictionary <string, string[]> mapping1 = ReadMapping(allIds, files[addtlSource], selection);
foreach (string key in mapping1.Keys.Where(key => !mapping.ContainsKey(key)))
{
mapping.Add(key, mapping1[key]);
}
}
SplitIds(type, out textColInds, out catColInds, out numColInds);
catCols = new string[catColInds.Length][][];
for (int i = 0; i < catCols.Length; i++)
{
catCols[i] = new string[nrows][];
}
textCols = new string[textColInds.Length][];
for (int i = 0; i < textCols.Length; i++)
{
textCols[i] = new string[nrows];
}
numCols = new double[numColInds.Length][];
for (int i = 0; i < numCols.Length; i++)
{
numCols[i] = new double[nrows];
}
for (int i = 0; i < nrows; i++)
{
string[] ids = baseIds[i].Length > 0 ? baseIds[i].Split(';') : new string[0];
HashSet <string>[] catVals = new HashSet <string> [catCols.Length];
for (int j = 0; j < catVals.Length; j++)
{
catVals[j] = new HashSet <string>();
}
HashSet <string>[] textVals = new HashSet <string> [textCols.Length];
for (int j = 0; j < textVals.Length; j++)
{
textVals[j] = new HashSet <string>();
}
List <double>[] numVals = new List <double> [numCols.Length];
for (int j = 0; j < numVals.Length; j++)
{
numVals[j] = new List <double>();
}
foreach (string id in ids)
{
if (mapping.ContainsKey(id))
{
string[] values = mapping[id];
AddCatVals(ArrayUtils.SubArray(values, catColInds), catVals);
AddTextVals(ArrayUtils.SubArray(values, textColInds), textVals);
AddNumVals(ArrayUtils.SubArray(values, numColInds), numVals);
}
else if (id.Contains("-"))
{
string q = id.Substring(0, id.IndexOf('-'));
if (mapping.ContainsKey(q))
{
string[] values = mapping[q];
AddCatVals(ArrayUtils.SubArray(values, catColInds), catVals);
AddTextVals(ArrayUtils.SubArray(values, textColInds), textVals);
AddNumVals(ArrayUtils.SubArray(values, numColInds), numVals);
}
}
}
for (int j = 0; j < catVals.Length; j++)
{
string[] q = ArrayUtils.ToArray(catVals[j]);
Array.Sort(q);
catCols[j][i] = q;
}
for (int j = 0; j < textVals.Length; j++)
{
string[] q = ArrayUtils.ToArray(textVals[j]);
Array.Sort(q);
textCols[j][i] = StringUtils.Concat(";", q);
}
for (int j = 0; j < numVals.Length; j++)
{
numCols[j][i] = ArrayUtils.Median(numVals[j]);
}
}
return(true);
}
