public static void printMyBoxplot(string outputFile, List <string> phenotypes)
{
//block (intercept) REngine from printing to the Console
//we are just redirecting the output of it to some StringWriter
var stdOut = Console.Out;
Console.SetOut(new StringWriter());
//print boxplot function
rEngineInstance.engine.Evaluate(
@"boxplotTissueCharge <- function(phenoMetabVals, plotTitle, plotYlabel, signifSymb, signifSymbYlevel, signifSymbXlevel) {
p <- ggplot(data=phenoMetabVals) +
theme_bw(base_size=18) +
ggtitle(plotTitle) +
ylab(plotYlabel) +
geom_boxplot(aes(factor(phenoMetabVals[,1]), phenoMetabVals[,2])) +" +
geom_textLinesInGgplot2Function(phenotypes.Count) +
@"theme(legend.position=""none"",
axis.title.x = element_blank());
return (p);
};");
//keep track of the last charge value so that we know when to print the tissue names
string lastCharge = metaboliteLevels.List_SampleForTissueAndCharge.Select(x => x.Charge).Distinct().OrderBy(x => x).Last();
//keep track of the first tissue value so that we know when to print the charge names
string firstTissue = metaboliteLevels.List_SampleForTissueAndCharge.Select(x => x.Tissue).Distinct().OrderBy(x => x).First();
//open the pdf stream
rEngineInstance.engine.Evaluate(@"pdf(file=""" + outputFile.Replace("\\", "/") + @""", width=14, height=9)");
//assisting list of tuples
List <Tuple <string, double> > phenoMetabValPairs, pValPairs;
//loop over custom metabolite IDs
foreach (string mtblid in metaboliteLevels.List_SampleForTissueAndCharge.SelectMany(x => x.ListOfMetabolites).Select(x => x.mtbltDetails.In_customId).Distinct().OrderBy(x => x))
{
//extract all the metabolites with the mtblid custom metabolite ID
msMetabolite mtbl = metaboliteLevels.List_SampleForTissueAndCharge.SelectMany(x => x.ListOfMetabolites).First(x => x.mtbltDetails.In_customId == mtblid).mtbltDetails;
//initialize the list that will store the boxplots
rEngineInstance.engine.Evaluate("accumulateBoxplots <- list()");
//loop over charges
foreach (string charge in metaboliteLevels.List_SampleForTissueAndCharge.Select(x => x.Charge).Distinct().OrderBy(x => x))
{
//loop over tissues
foreach (string tissue in metaboliteLevels.List_SampleForTissueAndCharge.Select(x => x.Tissue).Distinct().OrderBy(x => x))
{
//set the plot title
rEngineInstance.engine.SetSymbol("boxPlotTitleAndYlabel", returnTitleAndYlabelCharacterVector(charge, lastCharge, tissue, firstTissue));
//if the metabolite has been detected for the given combination of tissue and charge then do the plot
if (metaboliteLevels.List_SampleForTissueAndCharge.Where(x => x.Tissue == tissue && x.Charge == charge && !publicVariables.excludedPhenotypes.Contains(x.Phenotype))
.SelectMany(x => x.ListOfMetabolites).Any(x => x.mtbltDetails.In_customId == mtbl.In_customId))
{
//initialize the assisting variable
phenoMetabValPairs = new List <Tuple <string, double> >();
pValPairs = new List <Tuple <string, double> >();
//loop over all the metabolites for tissue and charge and non-ignore phenotypes
//in order to fill in the assisting variables
//these variables serve for significance symbols in the plot and for where to plce it in the plot
foreach (sampleForTissueAndCharge sftac in metaboliteLevels.List_SampleForTissueAndCharge.Where(x => x.Tissue == tissue && x.Charge == charge && !publicVariables.excludedPhenotypes.Contains(x.Phenotype)))
{
foreach (sampleForTissueAndCharge.parentMetabolite tdm in sftac.ListOfMetabolites.Where(x => x.mtbltDetails.In_customId == mtbl.In_customId))
{
phenoMetabValPairs.Add(new Tuple <string, double>(sftac.Phenotype, tdm.mtbltVals.Imputed));
if (pValPairs.Count == 0)
{
for (int i = 0; i < phenotypes.Count; i++)
{
for (int j = (i + 1); j < phenotypes.Count; j++)
{
pValPairs.Add(new Tuple <string, double>(phenotypes[i].First() + "v" + phenotypes[j].First(),
tdm.mtbltDetails.ListOfStats.PairwiseTestPvalue.First(x => x.group1 == phenotypes[i] && x.group2 == phenotypes[j]).pairValue));
}
}
if (publicVariables.numberOfClasses != publicVariables.numberOfClassesValues.two)
{
pValPairs.Add(new Tuple <string, double>("", tdm.mtbltDetails.ListOfStats.MultiGroupPvalue));
}
}
}
}
//significance symbols matrix
rEngineInstance.engine.SetSymbol("signifSymbYlevel", returnSignificanceDataFrame(phenoMetabValPairs, pValPairs));
//sets the dataframe variable df in R
rEngineInstance.engine.SetSymbol("df", returnIEnurable(phenoMetabValPairs));
//do not plot anything to the console
//adds the boxplot in the list of plots
//stop not printing stuff in teh console
rEngineInstance.engine.Evaluate(@"pdf(NULL);
accumulateBoxplots[[length(accumulateBoxplots) + 1]] <-
boxplotTissueCharge(df, boxPlotTitleAndYlabel[1], boxPlotTitleAndYlabel[2], signifSymbYlevel[,1], signifSymbYlevel[,2], signifSymbYlevel[,3]);
dev.off();");
}
else //if the metabolite has not been detected for this combination of tissue and charge provide and empty plot
{
//empty plot: is defined in the initialization of rEngineInstance
rEngineInstance.engine.Evaluate("accumulateBoxplots[[length(accumulateBoxplots) + 1]] <- nullPlot(boxPlotTitleAndYlabel[1], boxPlotTitleAndYlabel[2])");
}
}
}
//do the plot
rEngineInstance.engine.Evaluate(@"" + printBoxPlotGrid(metaboliteLevels.List_SampleForTissueAndCharge.Select(x => x.Charge).Distinct().Count(), metaboliteLevels.List_SampleForTissueAndCharge.Select(x => x.Tissue).Distinct().Count()) +
@"p <- grid.text(""" + mtbl.In_customId + @"_" + mtbl.In_Name + @""", x=unit(130,""mm""), y=unit(225,""mm""), just=c(""left"", ""top""), gp = gpar(fontface=""bold"", fontsize=24, col=""blue""));
print(p);");
}
rEngineInstance.engine.Evaluate(@"dev.off()");
//Re-enable Console printings
Console.SetOut(stdOut);
}
public static void printMyScatterplot(string outputFile)
{
//block (intercept) REngine from printing to the Console
//we are just redirecting the output of it to some StringWriter
var stdOut = Console.Out;
Console.SetOut(new StringWriter());
rEngineInstance.engine.Evaluate(
@"scatterplotTissueCharge <- function(df, plotTitle, plotYlabel, coords, rcorrLabel) {
corrLine <- coef(lm(df$metab_vals ~ df$clindt_vals));
p <- ggplot(df) +
theme_bw(base_size = 18) +
ggtitle(paste(plotTitle)) +
ylab(plotYlabel) +
geom_point(aes(x = df$clindt_vals, y = df$metab_vals, color = factor(df$pheno))) +
geom_abline(intercept = corrLine[1], slope = corrLine[2]) +
geom_text(data = NULL, x = coords[1], y = coords[2], label = rcorrLabel) +
theme(legend.position = ""none"",
axis.title.x = element_blank());
return (p);
};");
//keep track of the last charge value so that we know when to print the tissue names
string lastCharge = metaboliteLevels.List_SampleForTissueAndCharge.Select(x => x.Charge).Distinct().OrderBy(x => x).Last();
//keep track of the first tissue value so that we know when to print the charge names
string firstTissue = metaboliteLevels.List_SampleForTissueAndCharge.Select(x => x.Tissue).Distinct().OrderBy(x => x).First();
//assisting list of tuples
List <rDataFrame> phenoMetabValPairs;
Tuple <double, double> corrPval;
corrValCoords _corrValCoord;
foreach (sampleForTissueAndCharge.sampleClinicalData sClinData in metaboliteLevels.List_SampleForTissueAndCharge.First().ListOfNumClinicalData)
{
if (sClinData.typeOf == sampleForTissueAndCharge.sampleClinicalData.type.categorical)
{
continue;
}
//open the pdf stream
rEngineInstance.engine.Evaluate(@"pdf(file=""" + outputFile.Replace("\\", "/") + sClinData.name + @".pdf"", width=14, height=9)");
//loop over custom metabolite IDs
foreach (string mtblid in metaboliteLevels.List_SampleForTissueAndCharge.SelectMany(x => x.ListOfMetabolites).Select(x => x.mtbltDetails.In_customId).Distinct().OrderBy(x => x))
{
//extract all the metabolites with the mtblid custom metabolite ID
msMetabolite mtbl = metaboliteLevels.List_SampleForTissueAndCharge.SelectMany(x => x.ListOfMetabolites).First(x => x.mtbltDetails.In_customId == mtblid).mtbltDetails;
//initialize the list that will store the boxplots
rEngineInstance.engine.Evaluate("accumulateScatterplots <- list()");
//loop over charges
foreach (string charge in metaboliteLevels.List_SampleForTissueAndCharge.Select(x => x.Charge).Distinct().OrderBy(x => x))
{
//loop over tissues
foreach (string tissue in metaboliteLevels.List_SampleForTissueAndCharge.Select(x => x.Tissue).Distinct().OrderBy(x => x))
{
//set the plot title
rEngineInstance.engine.SetSymbol("scatterPlotTitleAndYlabel", returnTitleAndYlabelCharacterVector(charge, lastCharge, tissue, firstTissue));
//if the metabolite has been detected for the given combination of tissue and charge then do the plot
if (metaboliteLevels.List_SampleForTissueAndCharge.Where(x => x.Tissue == tissue && x.Charge == charge && !publicVariables.excludedPhenotypes.Contains(x.Phenotype))
.SelectMany(x => x.ListOfMetabolites).Any(x => x.mtbltDetails.In_customId == mtbl.In_customId))
{
//initialize the assisting variable
phenoMetabValPairs = new List <rDataFrame>();
//loop over all the metabolites for tissue and charge and non-ignore phenotypes
//in order to fill in the assisting variables
//these variables serve for significance symbols in the plot and for where to plce it in the plot
foreach (sampleForTissueAndCharge sftac in metaboliteLevels.List_SampleForTissueAndCharge.Where(x => x.Tissue == tissue && x.Charge == charge && !publicVariables.excludedPhenotypes.Contains(x.Phenotype)))
{
phenoMetabValPairs.Add(new rDataFrame()
{
phenotype = sftac.Phenotype,
metabolite = sftac.ListOfMetabolites.First(x => x.mtbltDetails.In_customId == mtbl.In_customId).mtbltVals.Imputed,
clinical_data = sftac.ListOfNumClinicalData.First(x => x.name == sClinData.name).n_value
});
}
//retrieve correlation value and p-value between metabolite and clinical data
corrPval = metaboliteLevels.List_SampleForTissueAndCharge.First(x => x.Tissue == tissue && x.Charge == charge && !publicVariables.excludedPhenotypes.Contains(x.Phenotype))
.ListOfMetabolites.First(x => x.mtbltDetails.In_customId == mtbl.In_customId).mtbltDetails.ListOfStats.CorrelationValues.Where(x => x.clinical_data_name == sClinData.name)
.Select(x => new Tuple <double, double>(Math.Round(x.corr_value, 2), x.pValueUnadjust)).ToList().First();
//significance symbols and coordinates
_corrValCoord = returnSignificanceDataFrame(phenoMetabValPairs, corrPval);
rEngineInstance.engine.SetSymbol("xCorrCoord", _corrValCoord.xCoord);
rEngineInstance.engine.SetSymbol("yCorrCoord", _corrValCoord.yCoord);
rEngineInstance.engine.Evaluate("coords <- c(xCorrCoord, yCorrCoord)");
rEngineInstance.engine.SetSymbol("rcorrLabel", _corrValCoord.label);
//sets the dataframe variable df in R
CharacterVector pheno = rEngineInstance.engine.CreateCharacterVector(phenoMetabValPairs.Select(x => x.phenotype).ToArray());
rEngineInstance.engine.SetSymbol("pheno", pheno);
NumericVector clindt_vals = rEngineInstance.engine.CreateNumericVector(phenoMetabValPairs.Select(x => x.clinical_data).ToArray());
rEngineInstance.engine.SetSymbol("clindt_vals", clindt_vals);
NumericVector metab_vals = rEngineInstance.engine.CreateNumericVector(phenoMetabValPairs.Select(x => x.metabolite).ToArray());
rEngineInstance.engine.SetSymbol("metab_vals", metab_vals);
rEngineInstance.engine.Evaluate("df <- cbind.data.frame(metab_vals, clindt_vals, pheno)");
//do not plot anything to the console
//adds the boxplot in the list of plots
//stop not printing stuff in teh console
rEngineInstance.engine.Evaluate(@"pdf(NULL);
accumulateScatterplots[[length(accumulateScatterplots) + 1]] <-
scatterplotTissueCharge(df, scatterPlotTitleAndYlabel[1], scatterPlotTitleAndYlabel[2], coords, rcorrLabel);
dev.off();");
}
else //if the metabolite has not been detected for this combination of tissue and charge provide and empty plot
{
//empty plot: is defined in the initialization of rEngineInstance
rEngineInstance.engine.Evaluate("accumulateScatterplots[[length(accumulateScatterplots) + 1]] <- nullPlot(scatterPlotTitleAndYlabel[1], scatterPlotTitleAndYlabel[2])");
}
}
}
//do the plot
rEngineInstance.engine.Evaluate(@"" + printBoxPlotGrid(metaboliteLevels.List_SampleForTissueAndCharge.Select(x => x.Charge).Distinct().Count(), metaboliteLevels.List_SampleForTissueAndCharge.Select(x => x.Tissue).Distinct().Count()) +
@"p <- grid.text(""x-" + sClinData.name + @" | y-" + mtbl.In_customId + @"_" + mtbl.In_Name + @""", x=unit(10,""mm""), y=unit(225,""mm""), just=c(""left"", ""top""), gp = gpar(fontface=""bold"", fontsize=24, col=""blue""));
print(p);");
}
rEngineInstance.engine.Evaluate(@"dev.off()");
}
//Re-enable Console printings
Console.SetOut(stdOut);
}
private static List <msMetabolite> ReadInputMetabolitesFromDatabase(List <string> listOfCsvLines)
{
List <msMetabolite> listOfMetabolitesPerTissueAndCharge = new List <msMetabolite>();
msMetabolite msMetab;
bool addToList = false, isDuplicate = false;
for (int i = publicVariables.indexToStartFrom; i < listOfCsvLines.First().Split(publicVariables.breakCharInFile).Length; i++)
{
msMetab = new msMetabolite()
{
In_Index = i,
In_Name = listOfCsvLines.First().Split(publicVariables.breakCharInFile).ElementAt(i).Trim(),
In_Type = listOfCsvLines.ElementAt(1).Split(publicVariables.breakCharInFile).ElementAt(i).Trim(),
In_Formula = listOfCsvLines.ElementAt(2).Split(publicVariables.breakCharInFile).ElementAt(i).Replace(" ", ""),
In_Mass = Convert.ToDouble(listOfCsvLines.ElementAt(3).Split(publicVariables.breakCharInFile).ElementAt(i)),
In_Rt = Convert.ToDouble(listOfCsvLines.ElementAt(4).Split(publicVariables.breakCharInFile).ElementAt(i)),
In_customId = listOfCsvLines.ElementAt(5).Split(publicVariables.breakCharInFile).ElementAt(i),
In_Cas_id = listOfCsvLines.ElementAt(6).Split(publicVariables.breakCharInFile).ElementAt(i),
In_add_Cas_id = listOfCsvLines.ElementAt(7).Split(publicVariables.breakCharInFile).ElementAt(i).Split('|').ToList(),
In_Hmdb_id = listOfCsvLines.ElementAt(8).Split(publicVariables.breakCharInFile).ElementAt(i),
In_add_Hmdb_id = listOfCsvLines.ElementAt(9).Split(publicVariables.breakCharInFile).ElementAt(i).Split('|').ToList(),
In_Kegg_id = listOfCsvLines.ElementAt(10).Split(publicVariables.breakCharInFile).ElementAt(i),
In_add_Kegg_id = listOfCsvLines.ElementAt(11).Split(publicVariables.breakCharInFile).ElementAt(i).Split('|').ToList(),
In_Chebi_id = listOfCsvLines.ElementAt(12).Split(publicVariables.breakCharInFile).ElementAt(i),
In_add_Chebi_id = listOfCsvLines.ElementAt(13).Split(publicVariables.breakCharInFile).ElementAt(i).Split('|').ToList(),
In_Pubchem_id = listOfCsvLines.ElementAt(14).Split(publicVariables.breakCharInFile).ElementAt(i),
In_add_Pubchem_id = listOfCsvLines.ElementAt(15).Split(publicVariables.breakCharInFile).ElementAt(i).Split('|').ToList(),
In_Chemspider_id = listOfCsvLines.ElementAt(16).Split(publicVariables.breakCharInFile).ElementAt(i),
In_add_Chemspider_id = listOfCsvLines.ElementAt(17).Split(publicVariables.breakCharInFile).ElementAt(i).Split('|').ToList(),
In_Lipidmaps_id = listOfCsvLines.ElementAt(18).Split(publicVariables.breakCharInFile).ElementAt(i),
In_add_Lipidmaps_id = listOfCsvLines.ElementAt(19).Split(publicVariables.breakCharInFile).ElementAt(i).Split('|').ToList(),
In_Metlin_id = listOfCsvLines.ElementAt(20).Split(publicVariables.breakCharInFile).ElementAt(i),
In_add_Metlin_id = listOfCsvLines.ElementAt(21).Split(publicVariables.breakCharInFile).ElementAt(i).Split('|').ToList(),
In_isProblematic = Convert.ToBoolean(listOfCsvLines.ElementAt(22).Split(publicVariables.breakCharInFile).ElementAt(i)),
In_msMsConfirmed = Convert.ToBoolean(listOfCsvLines.ElementAt(23).Split(publicVariables.breakCharInFile).ElementAt(i)),
In_inBlank = Convert.ToBoolean(listOfCsvLines.ElementAt(24).Split(publicVariables.breakCharInFile).ElementAt(i)),
In_msProblematic = Convert.ToBoolean(listOfCsvLines.ElementAt(25).Split(publicVariables.breakCharInFile).ElementAt(i)),
In_AZmSuperClass = listOfCsvLines.ElementAt(26).Split(publicVariables.breakCharInFile).ElementAt(i),
In_AZmClass = listOfCsvLines.ElementAt(27).Split(publicVariables.breakCharInFile).ElementAt(i),
In_AZmNameFixed = listOfCsvLines.ElementAt(28).Split(publicVariables.breakCharInFile).ElementAt(i)
};
if (!string.IsNullOrEmpty(msMetab.In_customId) && !string.IsNullOrWhiteSpace(msMetab.In_customId) && msMetab.In_Type == "Metabolite")
{
if (listOfMetaboliteIDs.Any(x => x.Split('_').First() == msMetab.In_customId))
{
if (listOfMetabolitesPerTissueAndCharge.Any(x => x.In_customId.Split('_').First() == msMetab.In_customId))
{
msMetab.ToHMDB_metabolite(listOfMetabolitesPerTissueAndCharge.First(x => x.In_customId.Split('_').First() == msMetab.In_customId));
if (listOfMetabolitesPerTissueAndCharge.Count(x => x.In_customId.Split('_').First() == msMetab.In_customId) == 1)
{
msMetab.In_customId = msMetab.In_customId + "_1";
}
else
{
msMetab.In_customId = msMetab.In_customId + "_" + Convert.ToString(listOfMetabolitesPerTissueAndCharge.Where(x => x.In_customId.Split('_').First() == msMetab.In_customId)
.Select(x => x.In_customId).Where(x => x.Split('_').Length > 1).Select(x => Convert.ToInt32(x.Split('_').Last())).Max() + 1);
}
addToList = true;
}
else
{
msMetab.ToHMDB_metabolite(List_SampleForTissueAndCharge.SelectMany(x => x.ListOfMetabolites).First(x => x.mtbltDetails.In_customId == msMetab.In_customId).mtbltDetails);
addToList = false;
}
isDuplicate = false;
}
else
{
msMetab.getFromMetaboliteDB();
addToList = true;
isDuplicate = false;
}
}
else if (msMetab.In_Type == "IS")
{
addToList = false;
isDuplicate = true;
}
if (addToList)
{
listOfMetaboliteIDs.Add(msMetab.In_customId);
}
if (!isDuplicate)
{
listOfMetabolitesPerTissueAndCharge.Add(msMetab);
}
}
return(listOfMetabolitesPerTissueAndCharge);
}
