org.fhcrc.cpl.viewer.ms2.commandline
Class PostProcessPepXMLCLM

java.lang.Object
  org.fhcrc.cpl.toolbox.commandline.BaseCommandLineModuleImpl
      org.fhcrc.cpl.viewer.commandline.modules.BaseViewerCommandLineModuleImpl
          org.fhcrc.cpl.viewer.ms2.commandline.PostProcessPepXMLCLM

All Implemented Interfaces:

CommandLineModule

public class PostProcessPepXMLCLM
extends BaseViewerCommandLineModuleImpl
implements CommandLineModule

post-process a pepxml file

Field Summary

protected static org.apache.log4j.Logger	_log
protected boolean	adjustQuantZeroAreas
protected java.lang.String	badProteinPrefix
protected boolean	excludeProteinPrefixQuantOnly
protected java.util.Map<java.io.File,java.lang.Float>	fileMedianLogRatioMap
protected java.util.Map<java.io.File,java.util.Map<java.lang.Integer,java.lang.Float>>	fileNumCysteinesMedianLogRatioMap
protected boolean	filterByProteinPrefix
protected java.lang.String	goodProteinPrefix
protected java.util.Set<java.lang.String>	heavyPeptidesAllRuns
protected java.lang.String[]	labelExplanations
protected java.lang.String[]	labelStrings
protected int	labelType
protected java.util.Set<java.lang.String>	lightPeptidesAllRuns
protected float	maxExpect
protected DeltaMassArgumentDefinition.DeltaMassWithType	maxFracDeltaMass
protected float	maxQuantExpect
protected boolean	medianCenter
protected boolean	medianCenterAllRunsTogether
protected boolean	medianCenterByNumCysteines
protected float	minPeptideProphet
protected float	minPeptideProphetForMedian
protected float	minQuantPeptideProphet
protected int	minRatiosForMedianCenter
protected java.io.File	outDir
protected java.io.File	outFile
protected java.util.Set<java.lang.String>	peptidesToStrip
protected java.io.File[]	pepXmlFiles
protected int	percentileForQuantZeroAreaAdjustment
protected java.util.Set<java.lang.String>	proteinsToKeep
protected java.util.Set<java.lang.String>	proteinsToStrip
protected boolean	requirePepXmlExtension
protected boolean	showCharts
protected boolean	stripLightIDs
protected boolean	stripQuantMissingLightOrHeavyAcrossAll
protected boolean	stripQuantMissingLightOrHeavyWithinRun
protected boolean	stripQuantNotInHeavyAcrossAll
protected boolean	stripQuantSingleScans
protected boolean	stripQuantZeroAreas

Fields inherited from class org.fhcrc.cpl.toolbox.commandline.BaseCommandLineModuleImpl

mArgumentDefs, mArgumentValues, mArgumentValueStrings, mCommandName, mHelpMessage, mShortDescription, mUsageMessage

Fields inherited from interface org.fhcrc.cpl.toolbox.commandline.CommandLineModule

MODULE_HELP_AUTOMATIC, MODULE_USAGE_AUTOMATIC, UNNAMED_ARG_SERIES_SEPARATOR

Constructor Summary

PostProcessPepXMLCLM()

Method Summary

protected void	addLightHeavyPeptides(FeatureSet featureSet)
protected void	adjustQuantZeroAreas(FeatureSet featureSet)
void	assignArgumentValues() the first step in invoking your module.
protected void	calcLogMedianRatiosAllFiles() TODO: fold this in with loadLightHeavyPeptides This is some really weird stuff, right here.
protected java.lang.String	calcOutputFilename(java.lang.String inputFilename)
protected int	countCysteines(java.lang.String peptide)
protected java.util.Map<java.lang.String,java.util.Map<java.lang.Integer,java.util.List<Feature>>>	createPeptideChargeFeatureListMap(FeatureSet featureSet)
void	execute() do the actual work
protected void	filterByProteinPrefix(FeatureSet featureSet) Does filtering by protein prefix.
protected void	filterOnQualityScores(FeatureSet featureSet)
protected void	handleFeatureFile(java.io.File featureFile, java.io.File outputFile)
protected void	init()
protected void	loadLightHeavyPeptidesAcrossAll()
protected double	log2(double input) Log base 2
protected java.util.List<java.lang.Float>	logMedianCenterOn0(java.util.List<java.lang.Float> allInputList, java.util.List<java.lang.Float> inputListForMedianCalc, java.lang.String chartTitleSuffix) median-centers the natural log on 0, not in place
protected void	processFeatureSet(FeatureSet featureSet)
protected java.util.List<java.lang.String>	readOneStringPerLine(java.io.File file) Read each line as a String, stopping at first whitespace
protected void	stripQuantWithoutLightOrHeavyIDWithinSet(FeatureSet featureSet)

Methods inherited from class org.fhcrc.cpl.viewer.commandline.modules.BaseViewerCommandLineModuleImpl

createUnnamedFeatureFileArgumentDefinition, createUnnamedSeriesFeatureFileArgumentDefinition, getFastaFileArgumentValue, getFeatureSetArgumentValue, getModificationListArgumentValue

Methods inherited from class org.fhcrc.cpl.toolbox.commandline.BaseCommandLineModuleImpl

addArgumentDefinition, addArgumentDefinitions, addArgumentDefinitions, assertArgumentAbsent, assertArgumentAbsent, assertArgumentPresent, assertArgumentPresent, createArgsTableHTML, createUnnamedFileArgumentDefinition, createUnnamedSeriesFileArgumentDefinition, digestArguments, getAdvancedArgumentDefinitions, getArgumentDefinition, getArgumentDefinitions, getArgumentDefinitionsSortedForDisplay, getArgumentValue, getArgumentValues, getArgumentValueStrings, getBasicArgumentDefinitions, getBooleanArgumentValue, getCommandName, getDeltaMassArgumentValue, getDoubleArgumentValue, getDoubleArrayArgumentValue, getFileArgumentValue, getFileArrayArgumentValue, getFloatArgumentValue, getFullHelp, getHelpMessage, getHtmlHelpFragment, getIntegerArgumentValue, getPrintWriter, getShortDescription, getStringArgumentValue, getUnnamedArgumentValue, getUnnamedFileArgumentValue, getUnnamedSeriesArgumentValues, getUnnamedSeriesFileArgumentValues, getUsage, hasAdvancedArguments, hasArgumentValue, hasUnnamedArgumentValue, hasUnnamedSeriesArgumentValue, invoke, makeHtmlSafe, sortArgDefsForDisplay, toString

Methods inherited from class java.lang.Object

clone, equals, finalize, getClass, hashCode, notify, notifyAll, wait, wait, wait

Methods inherited from interface org.fhcrc.cpl.toolbox.commandline.CommandLineModule

digestArguments, getAdvancedArgumentDefinitions, getArgumentDefinition, getArgumentDefinitions, getArgumentDefinitionsSortedForDisplay, getArgumentValueStrings, getBasicArgumentDefinitions, getCommandName, getFullHelp, getHelpMessage, getHtmlHelpFragment, getShortDescription, getUsage, invoke

Field Detail

_log

protected static org.apache.log4j.Logger _log

pepXmlFiles

protected java.io.File[] pepXmlFiles

medianCenter

protected boolean medianCenter

medianCenterByNumCysteines

protected boolean medianCenterByNumCysteines

medianCenterAllRunsTogether

protected boolean medianCenterAllRunsTogether

stripQuantMissingLightOrHeavyWithinRun

protected boolean stripQuantMissingLightOrHeavyWithinRun

stripQuantMissingLightOrHeavyAcrossAll

protected boolean stripQuantMissingLightOrHeavyAcrossAll

stripQuantNotInHeavyAcrossAll

protected boolean stripQuantNotInHeavyAcrossAll

stripLightIDs

protected boolean stripLightIDs

adjustQuantZeroAreas

protected boolean adjustQuantZeroAreas

stripQuantZeroAreas

protected boolean stripQuantZeroAreas

stripQuantSingleScans

protected boolean stripQuantSingleScans

fileMedianLogRatioMap

protected java.util.Map<java.io.File,java.lang.Float> fileMedianLogRatioMap

fileNumCysteinesMedianLogRatioMap

protected java.util.Map<java.io.File,java.util.Map<java.lang.Integer,java.lang.Float>> fileNumCysteinesMedianLogRatioMap

filterByProteinPrefix

protected boolean filterByProteinPrefix

percentileForQuantZeroAreaAdjustment

protected int percentileForQuantZeroAreaAdjustment

badProteinPrefix

protected java.lang.String badProteinPrefix

goodProteinPrefix

protected java.lang.String goodProteinPrefix

excludeProteinPrefixQuantOnly

protected boolean excludeProteinPrefixQuantOnly

outFile

protected java.io.File outFile

outDir

protected java.io.File outDir

peptidesToStrip

protected java.util.Set<java.lang.String> peptidesToStrip

proteinsToStrip

protected java.util.Set<java.lang.String> proteinsToStrip

proteinsToKeep

protected java.util.Set<java.lang.String> proteinsToKeep

showCharts

protected boolean showCharts

minRatiosForMedianCenter

protected int minRatiosForMedianCenter

minPeptideProphetForMedian

protected float minPeptideProphetForMedian

minPeptideProphet

protected float minPeptideProphet

minQuantPeptideProphet

protected float minQuantPeptideProphet

maxExpect

protected float maxExpect

maxQuantExpect

protected float maxQuantExpect

requirePepXmlExtension

protected boolean requirePepXmlExtension

maxFracDeltaMass

protected DeltaMassArgumentDefinition.DeltaMassWithType maxFracDeltaMass

labelType

protected int labelType

lightPeptidesAllRuns

protected java.util.Set<java.lang.String> lightPeptidesAllRuns

heavyPeptidesAllRuns

protected java.util.Set<java.lang.String> heavyPeptidesAllRuns

labelStrings

protected java.lang.String[] labelStrings

labelExplanations

protected java.lang.String[] labelExplanations

Constructor Detail

PostProcessPepXMLCLM

public PostProcessPepXMLCLM()

Method Detail

init

protected void init()

assignArgumentValues

public void assignArgumentValues()
                          throws ArgumentValidationException

Description copied from interface: CommandLineModule

the first step in invoking your module. The values assigned to the various arguments by the user are passed to your module for storage and additional validation. Any communication with the user about their argument values should be done by this method.

Specified by:

assignArgumentValues in interface CommandLineModule

Throws:

ArgumentValidationException

readOneStringPerLine

protected java.util.List<java.lang.String> readOneStringPerLine(java.io.File file)
                                                         throws ArgumentValidationException

Read each line as a String, stopping at first whitespace

Parameters:

file -

Returns:

Throws:

ArgumentValidationException

execute

public void execute()
             throws CommandLineModuleExecutionException

do the actual work

Specified by:

execute in interface CommandLineModule

Throws:

CommandLineModuleExecutionException

calcLogMedianRatiosAllFiles

protected void calcLogMedianRatiosAllFiles()
                                    throws CommandLineModuleExecutionException

TODO: fold this in with loadLightHeavyPeptides This is some really weird stuff, right here. This method handles calculation of median log ratios, regardless of whether you're doing it separately by file or all together, by number of cysteines or not. It creates a map data structure that gets used down below, with a key for each file. If we're doing global centering, the value gets repeated for each key. This is a bit silly, but it makes things simpler below.

Throws:

CommandLineModuleExecutionException

loadLightHeavyPeptidesAcrossAll

protected void loadLightHeavyPeptidesAcrossAll()
                                        throws CommandLineModuleExecutionException

Throws:

CommandLineModuleExecutionException

calcOutputFilename

protected java.lang.String calcOutputFilename(java.lang.String inputFilename)

handleFeatureFile

protected void handleFeatureFile(java.io.File featureFile,
                                 java.io.File outputFile)
                          throws CommandLineModuleExecutionException

Parameters:

featureFile -

outputFile -

Throws:

CommandLineModuleExecutionException

filterOnQualityScores

protected void filterOnQualityScores(FeatureSet featureSet)

countCysteines

protected int countCysteines(java.lang.String peptide)

processFeatureSet

protected void processFeatureSet(FeatureSet featureSet)

adjustQuantZeroAreas

protected void adjustQuantZeroAreas(FeatureSet featureSet)

filterByProteinPrefix

protected void filterByProteinPrefix(FeatureSet featureSet)

Does filtering by protein prefix. if excludeProteinPrefixQuantOnly is true, then this "filtering" just means removing quantitation

Parameters:

featureSet -

createPeptideChargeFeatureListMap

protected java.util.Map<java.lang.String,java.util.Map<java.lang.Integer,java.util.List<Feature>>> createPeptideChargeFeatureListMap(FeatureSet featureSet)

addLightHeavyPeptides

protected void addLightHeavyPeptides(FeatureSet featureSet)

stripQuantWithoutLightOrHeavyIDWithinSet

protected void stripQuantWithoutLightOrHeavyIDWithinSet(FeatureSet featureSet)

log2

protected double log2(double input)

Log base 2

Parameters:

input -

Returns:

logMedianCenterOn0

protected java.util.List<java.lang.Float> logMedianCenterOn0(java.util.List<java.lang.Float> allInputList,
                                                             java.util.List<java.lang.Float> inputListForMedianCalc,
                                                             java.lang.String chartTitleSuffix)

median-centers the natural log on 0, not in place

Parameters:

allInputList -

inputListForMedianCalc - this gets munged

Returns: