zzh
/
gatk-3.8
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity

Merge branch 'master' of ssh://nickel.broadinstitute.org/humgen/gsa-scr1/gsa-engineering/git/unstable

This commit is contained in:
Eric Banks 2011-08-19 14:13:43 -04:00
parent ee00171b6d 7b5fa4486d
commit e2c066ac1e
3 changed files with 218 additions and 104 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

33
public/java/src/org/broadinstitute/sting/gatk/walkers/validation/GenotypeAndValidateWalker.java
View File

@ -25,10 +25,7 @@
package org.broadinstitute.sting.gatk.walkers.validation;
import org.broadinstitute.sting.commandline.Argument;
import org.broadinstitute.sting.commandline.Input;
import org.broadinstitute.sting.commandline.Output;
import org.broadinstitute.sting.commandline.RodBinding;
import org.broadinstitute.sting.commandline.*;
import org.broadinstitute.sting.gatk.contexts.AlignmentContext;
import org.broadinstitute.sting.gatk.contexts.ReferenceContext;
import org.broadinstitute.sting.gatk.refdata.RefMetaDataTracker;
@ -201,30 +198,58 @@ import static org.broadinstitute.sting.utils.IndelUtils.isInsideExtendedIndel;
public class GenotypeAndValidateWalker extends RodWalker<GenotypeAndValidateWalker.CountedData, GenotypeAndValidateWalker.CountedData> implements TreeReducible<GenotypeAndValidateWalker.CountedData> {
/**
* The optional output file that will have all the variants used in the Genotype and Validation essay.
*/
@Output(doc="Generate a VCF file with the variants considered by the walker, with a new annotation \"callStatus\" which will carry the value called in the validation VCF or BAM file", required=false)
protected VCFWriter vcfWriter = null;
/**
* The callset to be used as truth (default) or validated (if BAM file is set to truth).
*/
@Input(fullName="alleles", shortName = "alleles", doc="The set of alleles at which to genotype", required=true)
public RodBinding<VariantContext> alleles;
/**
* Makes the Unified Genotyper calls to the BAM file the truth dataset and validates the alleles ROD binding callset.
*/
@Argument(fullName ="set_bam_truth", shortName ="bt", doc="Use the calls on the reads (bam file) as the truth dataset and validate the calls on the VCF", required=false)
private boolean bamIsTruth = false;
/**
* The minimum base quality score necessary for a base to be considered when calling a genotype. This argument is passed to the Unified Genotyper.
*/
@Argument(fullName="minimum_base_quality_score", shortName="mbq", doc="Minimum base quality score for calling a genotype", required=false)
private int mbq = -1;
/**
* The maximum deletion fraction allowed in a site for calling a genotype. This argument is passed to the Unified Genotyper.
*/
@Argument(fullName="maximum_deletion_fraction", shortName="deletions", doc="Maximum deletion fraction for calling a genotype", required=false)
private double deletions = -1;
/**
* the minimum phred-scaled Qscore threshold to separate high confidence from low confidence calls. This argument is passed to the Unified Genotyper.
*/
@Argument(fullName="standard_min_confidence_threshold_for_calling", shortName="stand_call_conf", doc="the minimum phred-scaled Qscore threshold to separate high confidence from low confidence calls", required=false)
private double callConf = -1;
/**
* the minimum phred-scaled Qscore threshold to emit low confidence calls. This argument is passed to the Unified Genotyper.
*/
@Argument(fullName="standard_min_confidence_threshold_for_emitting", shortName="stand_emit_conf", doc="the minimum phred-scaled Qscore threshold to emit low confidence calls", required=false)
private double emitConf = -1;
/**
* Only validate sites that have at least a given depth
*/
@Argument(fullName="condition_on_depth", shortName="depth", doc="Condition validation on a minimum depth of coverage by the reads", required=false)
private int minDepth = -1;
/**
* If your VCF or BAM file has more than one sample and you only want to validate one, use this parameter to choose it.
*/
@Hidden
@Argument(fullName ="sample", shortName ="sn", doc="Name of the sample to validate (in case your VCF/BAM has more than one sample)", required=false)
private String sample = "";

26
public/java/src/org/broadinstitute/sting/utils/help/GATKDoclet.java
View File

@ -34,7 +34,10 @@ import org.apache.commons.io.FileUtils;
import org.apache.log4j.Level;
import org.apache.log4j.Logger;
import org.broad.tribble.FeatureCodec;
import org.broadinstitute.sting.gatk.CommandLineGATK;
import org.broadinstitute.sting.gatk.walkers.qc.DocumentationTest;
import org.broadinstitute.sting.utils.exceptions.ReviewedStingException;
import org.broadinstitute.sting.utils.exceptions.UserException;
import java.io.*;
import java.util.*;
@ -48,6 +51,7 @@ public class GATKDoclet {
final protected static Logger logger = Logger.getLogger(GATKDoclet.class);
protected static String buildTimestamp = null, absoluteVersion = null;
protected static boolean showHiddenFeatures = false;
protected static boolean testOnly = false;
RootDoc rootDoc;
@ -75,6 +79,8 @@ public class GATKDoclet {
absoluteVersion = options[1];
if (options[0].equals("-include-hidden"))
showHiddenFeatures = true;
if (options[0].equals("-test"))
testOnly = true;
}
GATKDoclet doclet = new GATKDoclet();
@ -88,16 +94,26 @@ public class GATKDoclet {
* @return Number of potential parameters; 0 if not supported.
*/
public static int optionLength(String option) {
if(option.equals("-build-timestamp") || option.equals("-absolute-version") || option.equals("-include-hidden")) {
if(option.equals("-build-timestamp") ||
option.equals("-absolute-version") ||
option.equals("-include-hidden")) {
return 2;
}
return 0;
} else if ( option.equals("-test") )
return 1;
else
return 0;
}
public boolean showHiddenFeatures() {
return showHiddenFeatures;
}
public static boolean testOnly() {
return testOnly;
}
private static final List<Class<?>> testOnlyKeepers = Arrays.asList(
DocumentationTest.class, CommandLineGATK.class, UserException.class);
public Set<GATKDocWorkUnit> workUnits() {
TreeSet<GATKDocWorkUnit> m = new TreeSet<GATKDocWorkUnit>();
@ -105,6 +121,10 @@ public class GATKDoclet {
//logger.debug("Considering " + doc);
Class clazz = getClassForClassDoc(doc);
// don't add anything that's not DocumentationTest if we are in test mode
if ( clazz != null && testOnly && ! testOnlyKeepers.contains(clazz) )
continue;
//if ( clazz != null && clazz.getName().equals("org.broadinstitute.sting.gatk.walkers.annotator.AlleleBalance"))
// logger.debug("foo");

263
public/java/src/org/broadinstitute/sting/utils/help/GenericDocumentationHandler.java
View File

@ -24,6 +24,7 @@
package org.broadinstitute.sting.utils.help;
import com.google.java.contract.Ensures;
import com.google.java.contract.Requires;
import com.sun.javadoc.ClassDoc;
import com.sun.javadoc.FieldDoc;
@ -31,8 +32,10 @@ import com.sun.javadoc.RootDoc;
import com.sun.javadoc.Tag;
import org.apache.log4j.Logger;
import org.broad.tribble.Feature;
import org.broad.tribble.bed.FullBEDFeature;
import org.broadinstitute.sting.commandline.*;
import org.broadinstitute.sting.gatk.CommandLineGATK;
import org.broadinstitute.sting.gatk.arguments.DbsnpArgumentCollection;
import org.broadinstitute.sting.gatk.refdata.tracks.FeatureManager;
import org.broadinstitute.sting.utils.Utils;
import org.broadinstitute.sting.utils.classloader.JVMUtils;
@ -49,14 +52,18 @@ import java.util.*;
*/
public class GenericDocumentationHandler extends DocumentedGATKFeatureHandler {
private static Logger logger = Logger.getLogger(GenericDocumentationHandler.class);
GATKDocWorkUnit toProcess;
ClassDoc classdoc;
Set<GATKDocWorkUnit> all;
RootDoc rootDoc;
/** The Class we are documenting */
private GATKDocWorkUnit toProcess;
/** The set of all classes we are documenting, for cross-referencing */
private Set<GATKDocWorkUnit> all;
/** The JavaDoc root */
private RootDoc rootDoc;
@Override
public boolean includeInDocs(ClassDoc doc) {
// return true;
try {
Class type = HelpUtils.getClassForDoc(doc);
return JVMUtils.isConcrete(type);
@ -76,7 +83,6 @@ public class GenericDocumentationHandler extends DocumentedGATKFeatureHandler {
this.rootDoc = rootDoc;
this.toProcess = toProcessArg;
this.all = allArg;
this.classdoc = toProcess.classDoc;
//System.out.printf("%s class %s%n", toProcess.group, toProcess.classDoc);
Map<String, Object> root = new HashMap<String, Object>();
@ -88,71 +94,76 @@ public class GenericDocumentationHandler extends DocumentedGATKFeatureHandler {
toProcess.setHandlerContent((String)root.get("summary"), root);
}
/**
* Add high-level summary information about toProcess to root, such as its
* name, summary, description, version, etc.
*
* @param root
*/
protected void addHighLevelBindings(Map<String, Object> root) {
root.put("name", classdoc.name());
root.put("name", toProcess.classDoc.name());
// Extract overrides from the doc tags.
StringBuilder summaryBuilder = new StringBuilder();
for(Tag tag: classdoc.firstSentenceTags())
for(Tag tag: toProcess.classDoc.firstSentenceTags())
summaryBuilder.append(tag.text());
root.put("summary", summaryBuilder.toString());
root.put("description", classdoc.commentText().substring(summaryBuilder.toString().length()));
root.put("description", toProcess.classDoc.commentText().substring(summaryBuilder.toString().length()));
root.put("timestamp", toProcess.buildTimestamp);
root.put("version", toProcess.absoluteVersion);
for(Tag tag: classdoc.tags()) {
for(Tag tag: toProcess.classDoc.tags()) {
root.put(tag.name(), tag.text());
}
}
/**
* Add bindings describing related GATK capabilites to toProcess
* @param root
*/
protected void addRelatedBindings(Map<String, Object> root) {
List<Map<String, Object>> extraDocsData = new ArrayList<Map<String, Object>>();
// add in all of the explicitly related items
for ( final Class extraDocClass : toProcess.annotation.extraDocs() ) {
final GATKDocWorkUnit otherUnit = GATKDoclet.findWorkUnitForClass(extraDocClass, all);
if ( otherUnit == null )
throw new ReviewedStingException("Requested extraDocs for class without any documentation: " + extraDocClass);
extraDocsData.add(
new HashMap<String, Object>(){{
put("filename", otherUnit.filename);
put("name", otherUnit.name);}});
}
root.put("extradocs", extraDocsData);
}
/**
* Add information about all of the arguments available to toProcess to root
*
* @param root
*/
protected void addArgumentBindings(Map<String, Object> root) {
ParsingEngine parsingEngine = createStandardGATKParsingEngine();
// attempt to instantiate the class
Object instance = makeInstanceIfPossible(toProcess.clazz);
Map<String, List<Map<String, Object>>> args = new HashMap<String, List<Map<String, Object>>>();
Map<String, List<Map<String, Object>>> args = createArgumentMap();
root.put("arguments", args);
args.put("all", new ArrayList<Map<String, Object>>());
args.put("required", new ArrayList<Map<String, Object>>());
args.put("optional", new ArrayList<Map<String, Object>>());
args.put("advanced", new ArrayList<Map<String, Object>>());
args.put("hidden", new ArrayList<Map<String, Object>>());
args.put("depreciated", new ArrayList<Map<String, Object>>());
try {
for ( ArgumentSource argumentSource : parsingEngine.extractArgumentSources(HelpUtils.getClassForDoc(classdoc)) ) {
// loop over all of the arguments according to the parsing engine
for ( final ArgumentSource argumentSource : parsingEngine.extractArgumentSources(HelpUtils.getClassForDoc(toProcess.classDoc)) ) {
// todo -- why can you have multiple ones?
ArgumentDefinition argDef = argumentSource.createArgumentDefinitions().get(0);
FieldDoc fieldDoc = getFieldDoc(classdoc, argumentSource.field.getName());
Map<String, Object> argBindings = docForArgument(fieldDoc, argumentSource, argDef); // todo -- why can you have multiple ones?
FieldDoc fieldDoc = getFieldDoc(toProcess.classDoc, argumentSource.field.getName());
Map<String, Object> argBindings = docForArgument(fieldDoc, argumentSource, argDef);
if ( ! argumentSource.isHidden() || getDoclet().showHiddenFeatures() ) {
logger.debug(String.format("Processing %s", argumentSource));
String kind = "optional";
if ( argumentSource.isRequired() ) kind = "required";
else if ( argumentSource.isAdvanced() ) kind = "advanced";
else if ( argumentSource.isHidden() ) kind = "hidden";
else if ( argumentSource.isDeprecated() ) kind = "depreciated";
final String kind = docKindOfArg(argumentSource);
// get the value of the field
if ( instance != null ) {
Object value = getFieldValue(toProcess.clazz, instance, fieldDoc.name());
if ( value == null && argumentSource.createsTypeDefault() ) {
// handle the case where there's an implicit default
try {
value = argumentSource.typeDefaultDocString();
} catch (ReviewedStingException e) {
; // failed to create type default, don't worry about it
}
}
if ( value != null )
argBindings.put("defaultValue", prettyPrintValueString(value));
}
final Object value = argumentValue(toProcess.clazz, argumentSource);
if ( value != null )
argBindings.put("defaultValue", prettyPrintValueString(value));
args.get(kind).add(argBindings);
args.get("all").add(argBindings);
} else {
logger.debug(String.format("Skipping hidden feature %s", argumentSource));
}
}
@ -165,11 +176,78 @@ public class GenericDocumentationHandler extends DocumentedGATKFeatureHandler {
}
}
/**
* Return the argument kind (required, advanced, hidden, etc) of this argumentSource
* @param argumentSource
* @return
*/
@Requires("argumentSource != null")
@Ensures("result != null")
private String docKindOfArg(ArgumentSource argumentSource) {
if ( argumentSource.isRequired() ) return "required";
else if ( argumentSource.isAdvanced() ) return "advanced";
else if ( argumentSource.isHidden() ) return "hidden";
else if ( argumentSource.isDeprecated() ) return "depreciated";
else return "optional";
}
/**
* Attempts to determine the value of argumentSource in an instantiated version of c
* @param c
* @param argumentSource
* @return value of argumentSource, or null if this isn't possible
*/
@Requires({"c != null", "argumentSource != null"})
private Object argumentValue(Class c, ArgumentSource argumentSource) {
// get the value of the field
// attempt to instantiate the class
final Object instance = makeInstanceIfPossible(toProcess.clazz);
if ( instance != null ) {
final Object value = getFieldValue(instance, argumentSource.field.getName());
if ( value != null )
return value;
if ( argumentSource.createsTypeDefault() ) {
try { // handle the case where there's an implicit default
return argumentSource.typeDefaultDocString();
} catch (ReviewedStingException e) {
; // failed to create type default, don't worry about it
}
}
}
return null;
}
/**
* Create the argument map for holding class arguments
* @return
*/
private Map<String, List<Map<String, Object>>> createArgumentMap() {
Map<String, List<Map<String, Object>>> args = new HashMap<String, List<Map<String, Object>>>();
args.put("all", new ArrayList<Map<String, Object>>());
args.put("required", new ArrayList<Map<String, Object>>());
args.put("optional", new ArrayList<Map<String, Object>>());
args.put("advanced", new ArrayList<Map<String, Object>>());
args.put("hidden", new ArrayList<Map<String, Object>>());
args.put("depreciated", new ArrayList<Map<String, Object>>());
return args;
}
/**
* Sorts the individual argument list in unsorted according to CompareArgumentsByName
* @param unsorted
* @return
*/
private List<Map<String, Object>> sortArguments(List<Map<String, Object>> unsorted) {
Collections.sort(unsorted, new CompareArgumentsByName());
return unsorted;
}
/**
* Sort arguments by case-insensitive comparison ignoring the -- and - prefixes
*/
private class CompareArgumentsByName implements Comparator<Map<String, Object>> {
public int compare(Map<String, Object> x, Map<String, Object> y) {
return elt(x).compareTo(elt(y));
@ -186,25 +264,32 @@ public class GenericDocumentationHandler extends DocumentedGATKFeatureHandler {
}
}
private Object getFieldValue(Class c, Object instance, String fieldName) {
Field field = JVMUtils.findField(c, fieldName);
if ( field != null ) {
Object value = JVMUtils.getFieldValue(field, instance);
//System.out.printf("Fetched value of field %s in class %s: %s%n", fieldName, c, value);
return value;
} else {
return findFieldValueInArgumentCollections(c, instance, fieldName);
}
}
private Object findFieldValueInArgumentCollections(Class c, Object instance, String fieldName) {
for ( Field field : JVMUtils.getAllFields(c) ) {
/**
* Utility function that finds the value of fieldName in any fields of ArgumentCollection fields in
* instance of class c.
*
* @param instance the object to query for the field value
* @param fieldName the name of the field we are looking for in instance
* @return The value assigned to field in the ArgumentCollection, otherwise null
*/
private Object getFieldValue(Object instance, String fieldName) {
//
// subtle note. If you have a field named X that is an ArgumentCollection that
// contains a field X as well, you need only consider fields in the argumentCollection, not
// matching the argument itself.
//
// @ArgumentCollection
// protected DbsnpArgumentCollection dbsnp = new DbsnpArgumentCollection();
//
for ( Field field : JVMUtils.getAllFields(instance.getClass()) ) {
if ( field.isAnnotationPresent(ArgumentCollection.class) ) {
//System.out.printf("Searching for %s in argument collection field %s%n", fieldName, field);
Object fieldValue = JVMUtils.getFieldValue(field, instance);
Object value = getFieldValue(fieldValue.getClass(), fieldValue, fieldName);
Object value = getFieldValue(fieldValue, fieldName);
if ( value != null )
return value;
} else if ( field.getName().equals(fieldName) ) {
return JVMUtils.getFieldValue(field, instance);
}
}
@ -212,6 +297,8 @@ public class GenericDocumentationHandler extends DocumentedGATKFeatureHandler {
}
/**
* Pretty prints value
*
* Assumes value != null
* @param value
* @return
@ -246,6 +333,11 @@ public class GenericDocumentationHandler extends DocumentedGATKFeatureHandler {
return value.toString();
}
/**
* Attempt to instantiate class c, if possible. Returns null if this proves impossible.
* @param c
* @return
*/
private Object makeInstanceIfPossible(Class c) {
Object instance = null;
try {
@ -265,47 +357,16 @@ public class GenericDocumentationHandler extends DocumentedGATKFeatureHandler {
// this last one is super dangerous, but some of these methods catch ClassNotFoundExceptions
// and rethrow then as RuntimeExceptions
catch (RuntimeException e) {}
// finally {
// if ( instance == null )
// logger.warn(String.format("Unable to create instance of class %s => %s", c, instance));
// }
return instance;
}
protected void addRelatedBindings(Map<String, Object> root) {
List<Map<String, Object>> extraDocsData = new ArrayList<Map<String, Object>>();
// add in all of the explicitly related items
for ( final Class extraDocClass : toProcess.annotation.extraDocs() ) {
final GATKDocWorkUnit otherUnit = GATKDoclet.findWorkUnitForClass(extraDocClass, all);
if ( otherUnit == null )
throw new ReviewedStingException("Requested extraDocs for class without any documentation: " + extraDocClass);
extraDocsData.add(
new HashMap<String, Object>(){{
put("filename", otherUnit.filename);
put("name", otherUnit.name);}});
}
root.put("extradocs", extraDocsData);
}
private static final String classRelationship(Class me, Class other) {
if ( other.equals(me) )
// no circular references
return null;
else if ( other.isAssignableFrom(me) )
// toProcess is a superclass of other.clazz
return "superclass";
else if ( me.isAssignableFrom(other) )
// toProcess inherits from other.clazz
return "subclass";
else
return null;
}
protected ParsingEngine createStandardGATKParsingEngine() {
/**
* Create an instance of the GATK parsing engine, for argument processing with GATKDoclet
* @return
*/
private ParsingEngine createStandardGATKParsingEngine() {
CommandLineProgram clp = new CommandLineGATK();
try {
CommandLineProgram.start(clp, new String[]{}, true);
@ -315,6 +376,14 @@ public class GenericDocumentationHandler extends DocumentedGATKFeatureHandler {
}
}
/**
* Gets the javadocs associated with field name in classDoc. Throws a
* runtime exception if this proves impossible.
*
* @param classDoc
* @param name
* @return
*/
private FieldDoc getFieldDoc(ClassDoc classDoc, String name) {
return getFieldDoc(classDoc, name, true);
}