public static int InferInnerDistance(string spritzDirectory, string analysisDirectory, string bamPath, string geneModelPath, out string[] outputFiles)
{
if (Path.GetExtension(geneModelPath) != ".bed")
{
geneModelPath = BEDOPSWrapper.GffOrGtf2Bed12(spritzDirectory, analysisDirectory, geneModelPath);
}
outputFiles = new string[]
{
Path.Combine(Path.GetDirectoryName(bamPath), Path.GetFileNameWithoutExtension(bamPath)) + InnerDistanceRPlotSuffix,
Path.Combine(Path.GetDirectoryName(bamPath), Path.GetFileNameWithoutExtension(bamPath)) + InnerDistanceFrequencyTableSuffix,
Path.Combine(Path.GetDirectoryName(bamPath), Path.GetFileNameWithoutExtension(bamPath)) + InnerDistanceDistanceTableSuffix
};
WrapperUtility.GenerateAndRunScript(WrapperUtility.GetAnalysisScriptPath(analysisDirectory, "InferInnerDistance.bash"), new List <string>
{
WrapperUtility.ChangeToToolsDirectoryCommand(spritzDirectory),
"python RSeQC-2.6.4/scripts/inner_distance.py" +
" -i " + WrapperUtility.ConvertWindowsPath(bamPath) + // input
" -o " + WrapperUtility.ConvertWindowsPath(Path.Combine(Path.GetDirectoryName(bamPath), Path.GetFileNameWithoutExtension(bamPath))) + // out prefix
" -r " + WrapperUtility.ConvertWindowsPath(geneModelPath), // gene model in BED format
WrapperUtility.EnsureClosedFileCommands(outputFiles[0]),
WrapperUtility.EnsureClosedFileCommands(outputFiles[1]),
WrapperUtility.EnsureClosedFileCommands(outputFiles[2]),
}).WaitForExit();
string[] distance_lines = File.ReadAllLines(Path.Combine(Path.GetDirectoryName(bamPath), Path.GetFileNameWithoutExtension(bamPath)) + InnerDistanceDistanceTableSuffix);
List <int> distances = new List <int>();
foreach (string dline in distance_lines)
{
if (int.TryParse(dline.Split('\t')[1], out int distance) &&
distance <250 && distance> -250) // default settings for infer_distance
{
distances.Add(distance);
}
}
int averageDistance = (int)Math.Round(distances.Average(), 0);
return(averageDistance);
}
/// <summary>
/// Annotate predicted transcripts
/// </summary>
/// <param name="spritzDirectory"></param>
/// <param name="threads"></param>
/// <param name="predictedGeneModelUcscBedPath"></param>
/// <param name="reference"></param>
/// <param name="slnckyOutPrefix"></param>
/// <returns></returns>
public static List <string> Annotate(string spritzDirectory, string analysisDirectory, int threads, string predictedGeneModelGtfPath, string reference, string slnckyOutPrefix)
{
string sortedBed12Cuffmerge = BEDOPSWrapper.GffOrGtf2Bed12(spritzDirectory, analysisDirectory, predictedGeneModelGtfPath);
string ucscCuffmergeBedPath = EnsemblDownloadsWrapper.ConvertFirstColumnEnsembl2UCSC(spritzDirectory, reference, sortedBed12Cuffmerge);
Directory.CreateDirectory(Path.GetDirectoryName(slnckyOutPrefix));
string ucscReference = reference.Contains("38") ? "hg38" : "hg19";
return(new List <string>
{
WrapperUtility.ChangeToToolsDirectoryCommand(spritzDirectory),
"cd slncky",
"if [[ ! -f " + WrapperUtility.ConvertWindowsPath(slnckyOutPrefix + LncsBedSuffix) + " || ! -s " + WrapperUtility.ConvertWindowsPath(slnckyOutPrefix + LncsBedSuffix) + " ]]; then " +
"./slncky.v1.0" +
" --threads " + threads.ToString() +
" " + WrapperUtility.ConvertWindowsPath(ucscCuffmergeBedPath) +
" " + ucscReference +
" " + WrapperUtility.ConvertWindowsPath(slnckyOutPrefix) +
"; fi"
});
}
