/// <summary>
/// Transcript assembly. Note that fragment bias estimation (--frag-bias-correct) and multi-read rescuing (--multi-read-correct) are not used.
/// These take a lot of time, and they only provide better abundance estimates, which we use RSEM for.
/// </summary>
/// <param name="spritzDirectory"></param>
/// <param name="threads"></param>
/// <param name="bamPath"></param>
/// <param name="geneModelGtfOrGffPath"></param>
/// <param name="strandSpecific"></param>
/// <param name="inferStrandSpecificity"></param>
/// <param name="outputTranscriptGtfPath"></param>
public static List <string> AssembleTranscripts(string spritzDirectory, int threads, string bamPath, string geneModelGtfOrGffPath, Genome genome,
Strandedness strandSpecific, bool inferStrandSpecificity, out string outputTranscriptGtfPath)
{
Strandedness strandedness = strandSpecific;
if (inferStrandSpecificity)
{
BAMProperties bamProperties = new BAMProperties(bamPath, geneModelGtfOrGffPath, genome, 0.8);
strandedness = bamProperties.Strandedness;
}
string sortedCheckPath = Path.Combine(Path.GetDirectoryName(bamPath), Path.GetFileNameWithoutExtension(bamPath) + ".cufflinksSortCheck");
outputTranscriptGtfPath = Path.Combine(Path.GetDirectoryName(bamPath), Path.GetFileNameWithoutExtension(bamPath) + ".gtf");
return(new List <string>
{
WrapperUtility.ChangeToToolsDirectoryCommand(spritzDirectory),
"samtools view -H " + WrapperUtility.ConvertWindowsPath(bamPath) + " | grep SO:coordinate > " + WrapperUtility.ConvertWindowsPath(sortedCheckPath),
"if [ ! -s " + WrapperUtility.ConvertWindowsPath(sortedCheckPath) + " ]; then " + SamtoolsWrapper.SortBam(bamPath, threads) + "; fi",
"bam=" + WrapperUtility.ConvertWindowsPath(bamPath),
"if [ ! -s " + WrapperUtility.ConvertWindowsPath(sortedCheckPath) + " ]; then bam=" + WrapperUtility.ConvertWindowsPath(Path.Combine(Path.GetDirectoryName(bamPath), Path.GetFileNameWithoutExtension(bamPath) + ".sorted.bam")) + "; fi",
"if [[ ! -f " + WrapperUtility.ConvertWindowsPath(outputTranscriptGtfPath) + " || ! -s " + WrapperUtility.ConvertWindowsPath(outputTranscriptGtfPath) + " ]]; then",
" echo \"Performing stringtie transcript reconstruction on " + bamPath + "\"",
" stringtie $bam " +
" -p " + threads.ToString() +
" -G " + WrapperUtility.ConvertWindowsPath(geneModelGtfOrGffPath) +
" -o " + WrapperUtility.ConvertWindowsPath(outputTranscriptGtfPath) +
(strandedness == Strandedness.None ? "" : strandedness == Strandedness.Forward ? "--fr" : "--rf"),
"fi",
});
}
/// <summary>
/// Transcript assembly. Note that fragment bias estimation (--frag-bias-correct) and multi-read rescuing (--multi-read-correct) are not used.
/// These take a lot of time, and they only provide better abundance estimates, which we use RSEM for.
/// </summary>
/// <param name="spritzDirectory"></param>
/// <param name="threads"></param>
/// <param name="bamPath"></param>
/// <param name="geneModelGtfOrGffPath"></param>
/// <param name="strandSpecific"></param>
/// <param name="inferStrandSpecificity"></param>
/// <param name="outputDirectory"></param>
public static List <string> AssembleTranscripts(string spritzDirectory, string analysisDirectory, int threads, string bamPath, string geneModelGtfOrGffPath, Genome genome, bool strandSpecific, bool inferStrandSpecificity, out string outputDirectory)
{
bool isStranded = strandSpecific;
if (inferStrandSpecificity)
{
BAMProperties bamProperties = new BAMProperties(bamPath, geneModelGtfOrGffPath, genome, 0.8);
isStranded = bamProperties.Strandedness != Strandedness.None;
}
string sortedCheckPath = Path.Combine(Path.GetDirectoryName(bamPath), Path.GetFileNameWithoutExtension(bamPath) + ".cufflinksSortCheck");
outputDirectory = Path.Combine(Path.GetDirectoryName(bamPath), Path.GetFileNameWithoutExtension(bamPath) + ".cufflinksOutput");
string script_name = WrapperUtility.GetAnalysisScriptPath(analysisDirectory, "CufflinksRun.bash");
return(new List <string>
{
WrapperUtility.ChangeToToolsDirectoryCommand(spritzDirectory),
"samtools view -H " + WrapperUtility.ConvertWindowsPath(bamPath) + " | grep SO:coordinate > " + WrapperUtility.ConvertWindowsPath(sortedCheckPath),
"if [ ! -s " + WrapperUtility.ConvertWindowsPath(sortedCheckPath) + " ]; then " + SamtoolsWrapper.SortBam(bamPath, threads) + "; fi",
"bam=" + WrapperUtility.ConvertWindowsPath(bamPath),
"if [ ! -s " + WrapperUtility.ConvertWindowsPath(sortedCheckPath) + " ]; then bam=" + WrapperUtility.ConvertWindowsPath(Path.Combine(Path.GetDirectoryName(bamPath), Path.GetFileNameWithoutExtension(bamPath) + ".sorted.bam")) + "; fi",
"if [[ ! -f " + WrapperUtility.ConvertWindowsPath(Path.Combine(outputDirectory, TranscriptsFilename)) + " || ! -s " + WrapperUtility.ConvertWindowsPath(Path.Combine(outputDirectory, TranscriptsFilename)) + " ]]; then " +
"cufflinks-2.2.1/cufflinks " +
" --num-threads " + threads.ToString() +
" --GTF-guide " + WrapperUtility.ConvertWindowsPath(geneModelGtfOrGffPath) +
" --output-dir " + WrapperUtility.ConvertWindowsPath(outputDirectory) +
(isStranded ? "--library-type fr-firststrand" : "") +
" $bam" +
"; fi",
});
}
/// <summary>
/// Gets commands to calculate expression an RSEM reference
/// </summary>
/// <param name="spritzDirectory"></param>
/// <returns></returns>
public List <string> CalculateExpressionCommands(string spritzDirectory, string referencePrefix, int threads, RSEMAlignerOption aligner, Strandedness strandedness,
string[] fastqPaths, bool doOuptutBam)
{
if (fastqPaths.Length < 1)
{
throw new ArgumentOutOfRangeException("No fastq files were given for RSEM calculate expression.");
}
if (fastqPaths.Length > 2)
{
throw new ArgumentOutOfRangeException("Too many fastq file types given for RSEM calculate expression.");
}
List <string> scriptCommands = new List <string>
{
WrapperUtility.ChangeToToolsDirectoryCommand(spritzDirectory),
"cd RSEM-1.3.0",
};
string[] analysisFastqPaths = fastqPaths;
string alignerOption = GetAlignerOption(spritzDirectory, aligner);
string threadOption = "--num-threads " + threads.ToString();
string strandOption = "--strandedness " + strandedness.ToString().ToLowerInvariant();
// Decompress files if needed
// The '--star-gzipped-read-file' and '--star-bzipped-read-file' options work, but then the rest of RSEM doesn't when using compressed files...
bool fastqIsGunzipped = analysisFastqPaths[0].EndsWith("gz");
bool fastqIsBunzipped = analysisFastqPaths[0].EndsWith("bz2") || analysisFastqPaths[0].EndsWith("bz");
if (fastqIsGunzipped || fastqIsBunzipped)
{
for (int i = 0; i < analysisFastqPaths.Length; i++)
{
string decompressionCommand = fastqIsGunzipped ? "gunzip" : "bunzip2";
scriptCommands.Add($"{decompressionCommand} --keep {WrapperUtility.ConvertWindowsPath(analysisFastqPaths[i])}");
analysisFastqPaths[i] = Path.ChangeExtension(analysisFastqPaths[i], null);
}
}
string inputOption = analysisFastqPaths.Length == 1 ? string.Join(",", analysisFastqPaths[0].Split(',').Select(f => WrapperUtility.ConvertWindowsPath(f))) :
"--paired-end " +
string.Join(",", analysisFastqPaths[0].Split(',').Select(f => WrapperUtility.ConvertWindowsPath(f))) +
" " +
string.Join(",", analysisFastqPaths[1].Split(',').Select(f => WrapperUtility.ConvertWindowsPath(f)));
var megabytes = Math.Floor((double)Process.GetCurrentProcess().VirtualMemorySize64 / 1000000);
string bamOption = doOuptutBam ? "--output-genome-bam" : "--no-bam-output";
OutputPrefix = Path.Combine(Path.GetDirectoryName(analysisFastqPaths[0].Split(',')[0]),
Path.GetFileNameWithoutExtension(analysisFastqPaths[0].Split(',')[0]) +
"_" + Path.GetExtension(analysisFastqPaths[0].Split(',')[0]).Substring(1).ToUpperInvariant() +
referencePrefix.GetHashCode().ToString());
// RSEM likes to sort the transcript.bam file, which takes forever and isn't very useful, I've found. Just sort the genome.bam file instead
string samtoolsCommands = !doOuptutBam ?
"" :
"if [[ ! -f " + WrapperUtility.ConvertWindowsPath(OutputPrefix + GenomeSortedBamSuffix) + " && ! -s " + WrapperUtility.ConvertWindowsPath(OutputPrefix + GenomeSortedBamSuffix) + " ]]; then\n" +
" " + SamtoolsWrapper.SortBam(OutputPrefix + GenomeBamSuffix, threads) + "\n" +
" " + SamtoolsWrapper.IndexBamCommand(OutputPrefix + GenomeSortedBamSuffix) + "\n" +
"fi";
// construct the commands
scriptCommands.AddRange(new List <string>
{
"if [[ ! -f " + WrapperUtility.ConvertWindowsPath(OutputPrefix + IsoformResultsSuffix) + " && ! -s " + WrapperUtility.ConvertWindowsPath(OutputPrefix + IsoformResultsSuffix) + " ]]; then " +
"./rsem-calculate-expression " +
"--time " + // include timed results
"--calc-ci " + // posterior calculation of 95% confidence intervals
alignerOption + " " +
threadOption + " " +
bamOption + " " +
inputOption + " " +
WrapperUtility.ConvertWindowsPath(referencePrefix) + " " +
WrapperUtility.ConvertWindowsPath(OutputPrefix) +
"; fi",
samtoolsCommands
});
return(scriptCommands);
}