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

public/java/src/org/broadinstitute/sting/commandline/ArgumentTypeDescriptor.java

@@ -373,15 +373,15 @@ class RodBindingArgumentTypeDescriptor extends ArgumentTypeDescriptor {
                         if ( featureDescriptor != null ) {
                             tribbleType = featureDescriptor.getName();
                             logger.warn("Dynamically determined type of " + file + " to be " + tribbleType);
-                        }
+                        } else {
-                    }
-                }
-            }
-
-            if ( tribbleType == null ) // error handling
                             throw new UserException.CommandLineException(
-                        String.format("Could not parse argument %s with value %s",
+                                    String.format("No tribble type was provided on the command line and the type of the file could not be determined dynamically. " +
-                                defaultDefinition.fullName, value));
+                                            "Please add an explicit type tag :TYPE listing the correct type from among the supported types: %s",
+                                            manager.userFriendlyListOfAvailableFeatures()));
+                        }
+                    }
+                }
+            }
 
             Constructor ctor = (makeRawTypeIfNecessary(type)).getConstructor(Class.class, String.class, String.class, String.class, Tags.class);
             Class parameterType = getParameterizedTypeClass(type);
@@ -395,8 +395,8 @@ class RodBindingArgumentTypeDescriptor extends ArgumentTypeDescriptor {
                             value, source.field.getName()));
         } catch (Exception e) {
             throw new UserException.CommandLineException(
-                    String.format("Failed to parse value %s for argument %s.",
+                    String.format("Failed to parse value %s for argument %s. Message: %s",
-                            value, source.field.getName()));
+                            value, source.field.getName(), e.getMessage()));
         }
     }
 

public/java/src/org/broadinstitute/sting/gatk/refdata/features/refseq/RefSeqCodec.java

@@ -105,6 +105,6 @@ public class RefSeqCodec implements ReferenceDependentFeatureCodec<RefSeqFeature
 
     @Override
     public Class getFeatureType() {
-        return RefSeqCodec.class;
+        return RefSeqFeature.class;
     }
 }

public/java/src/org/broadinstitute/sting/gatk/walkers/annotator/SnpEff.java

@@ -161,19 +161,19 @@ public class SnpEff extends InfoFieldAnnotation implements ExperimentalAnnotatio
 
     public List<VCFInfoHeaderLine> getDescriptions() {
         return Arrays.asList(
-            new VCFInfoHeaderLine(GENE_ID_KEY,                  1, VCFHeaderLineType.String,  "Gene ID"),
+            new VCFInfoHeaderLine(GENE_ID_KEY,                  1, VCFHeaderLineType.String,  "Gene ID for the highest-impact effect resulting from the current variant"),
-            new VCFInfoHeaderLine(GENE_NAME_KEY,                1, VCFHeaderLineType.String,  "Gene name"),
+            new VCFInfoHeaderLine(GENE_NAME_KEY,                1, VCFHeaderLineType.String,  "Gene name for the highest-impact effect resulting from the current variant"),
-            new VCFInfoHeaderLine(TRANSCRIPT_ID_KEY,            1, VCFHeaderLineType.String,  "Transcript ID"),
+            new VCFInfoHeaderLine(TRANSCRIPT_ID_KEY,            1, VCFHeaderLineType.String,  "Transcript ID for the highest-impact effect resulting from the current variant"),
-            new VCFInfoHeaderLine(EXON_ID_KEY,                  1, VCFHeaderLineType.String,  "Exon ID"),
+            new VCFInfoHeaderLine(EXON_ID_KEY,                  1, VCFHeaderLineType.String,  "Exon ID for the highest-impact effect resulting from the current variant"),
-            new VCFInfoHeaderLine(EXON_RANK_KEY,                1, VCFHeaderLineType.Integer, "Exon rank"),
+            new VCFInfoHeaderLine(EXON_RANK_KEY,                1, VCFHeaderLineType.Integer, "Exon rank for the highest-impact effect resulting from the current variant"),
-            new VCFInfoHeaderLine(WITHIN_NON_CODING_GENE_KEY,   0, VCFHeaderLineType.Flag,    "If present, gene is non-coding"),
+            new VCFInfoHeaderLine(WITHIN_NON_CODING_GENE_KEY,   0, VCFHeaderLineType.Flag,    "If this flag is present, the highest-impact effect resulting from the current variant is within a non-coding gene"),
-            new VCFInfoHeaderLine(EFFECT_KEY,                   1, VCFHeaderLineType.String,  "One of the most high-impact effects across all transcripts at this site"),
+            new VCFInfoHeaderLine(EFFECT_KEY,                   1, VCFHeaderLineType.String,  "The highest-impact effect resulting from the current variant (or one of the highest-impact effects, if there is a tie)"),
-            new VCFInfoHeaderLine(EFFECT_IMPACT_KEY,            1, VCFHeaderLineType.String,  "Impact of the effect " + Arrays.toString(SnpEffConstants.EffectImpact.values())),
+            new VCFInfoHeaderLine(EFFECT_IMPACT_KEY,            1, VCFHeaderLineType.String,  "Impact of the highest-impact effect resulting from the current variant " + Arrays.toString(SnpEffConstants.EffectImpact.values())),
-            new VCFInfoHeaderLine(EFFECT_EXTRA_INFORMATION_KEY, 1, VCFHeaderLineType.String,  "Additional information about the effect"),
+            new VCFInfoHeaderLine(EFFECT_EXTRA_INFORMATION_KEY, 1, VCFHeaderLineType.String,  "Additional information about the highest-impact effect resulting from the current variant"),
-            new VCFInfoHeaderLine(OLD_NEW_AA_KEY,               1, VCFHeaderLineType.String,  "Old/New amino acid"),
+            new VCFInfoHeaderLine(OLD_NEW_AA_KEY,               1, VCFHeaderLineType.String,  "Old/New amino acid for the highest-impact effect resulting from the current variant"),
-            new VCFInfoHeaderLine(OLD_NEW_CODON_KEY,            1, VCFHeaderLineType.String,  "Old/New codon"),
+            new VCFInfoHeaderLine(OLD_NEW_CODON_KEY,            1, VCFHeaderLineType.String,  "Old/New codon for the highest-impact effect resulting from the current variant"),
-            new VCFInfoHeaderLine(CODON_NUM_KEY,                1, VCFHeaderLineType.Integer, "Codon number"),
+            new VCFInfoHeaderLine(CODON_NUM_KEY,                1, VCFHeaderLineType.Integer, "Codon number for the highest-impact effect resulting from the current variant"),
-            new VCFInfoHeaderLine(CDS_SIZE_KEY,                 1, VCFHeaderLineType.Integer, "CDS size")
+            new VCFInfoHeaderLine(CDS_SIZE_KEY,                 1, VCFHeaderLineType.Integer, "CDS size for the highest-impact effect resulting from the current variant")
         );
     }
 }

public/java/src/org/broadinstitute/sting/gatk/walkers/beagle/BeagleOutputToVCFWalker.java

@@ -48,6 +48,31 @@ import static java.lang.Math.log10;
 
 /**
  * Takes files produced by Beagle imputation engine and creates a vcf with modified annotations.
+ *
+ * <p>This walker is intended to be run after Beagle has successfully executed. The full calling sequence for using Beagle along with the GATK is:      </p>
+ *
+ * <p>1. Run ProduceBeagleInputWalker.  </p>
+ * <p>2. Run Beagle</p>
+ * <p>3. Uncompress output files</p>
+ * <p>4. Run BeagleOutputToVCFWalker.</p>
+ *
+ *
+ * Note that this walker requires all input files produced by Beagle.
+ *
+ *
+ * <h2>Example</h2>
+ * <pre>
+ *     java -Xmx4000m -jar dist/GenomeAnalysisTK.jar \
+ *      -R reffile.fasta -T BeagleOutputToVCF \
+ *      -V input_vcf.vcf \
+ *      -beagleR2:BEAGLE /myrun.beagle_output.r2 \
+ *      -beaglePhased:BEAGLE /myrun.beagle_output.phased \
+ *      -beagleProbs:BEAGLE /myrun.beagle_output.gprobs \
+ *      -o output_vcf.vcf
+ *      </pre>
+
+ <p> Note that Beagle produces some of these files compressed as .gz, so gunzip must be run on them before walker is run in order to decompress them </p>
+
  */
 public class BeagleOutputToVCFWalker  extends RodWalker<Integer, Integer> {
 
@@ -57,22 +82,18 @@ public class BeagleOutputToVCFWalker  extends RodWalker<Integer, Integer> {
     @Input(fullName="comp", shortName = "comp", doc="Comparison VCF file", required=false)
     public RodBinding<VariantContext> comp;
 
-    @Input(fullName="beagleR2", shortName = "beagleR2", doc="VCF file", required=true)
+    @Input(fullName="beagleR2", shortName = "beagleR2", doc="Beagle-produced .r2 file containing R^2 values for all markers", required=true)
     public RodBinding<BeagleFeature> beagleR2;
 
-    @Input(fullName="beagleProbs", shortName = "beagleProbs", doc="VCF file", required=true)
+    @Input(fullName="beagleProbs", shortName = "beagleProbs", doc="Beagle-produced .probs file containing posterior genotype probabilities", required=true)
     public RodBinding<BeagleFeature> beagleProbs;
 
-    @Input(fullName="beaglePhased", shortName = "beaglePhased", doc="VCF file", required=true)
+    @Input(fullName="beaglePhased", shortName = "beaglePhased", doc="Beagle-produced .phased file containing phased genotypes", required=true)
     public RodBinding<BeagleFeature> beaglePhased;
 
-    @Output(doc="File to which variants should be written",required=true)
+    @Output(doc="VCF File to which variants should be written",required=true)
     protected VCFWriter vcfWriter = null;
 
-    @Argument(fullName="output_file", shortName="output", doc="Please use --out instead" ,required=false)
-    @Deprecated
-    protected String oldOutputArg;
-
     @Argument(fullName="dont_mark_monomorphic_sites_as_filtered", shortName="keep_monomorphic", doc="If provided, we won't filter sites that beagle tags as monomorphic.  Useful for imputing a sample's genotypes from a reference panel" ,required=false)
     public boolean DONT_FILTER_MONOMORPHIC_SITES = false;
 

public/java/src/org/broadinstitute/sting/gatk/walkers/beagle/ProduceBeagleInputWalker.java

@@ -48,18 +48,44 @@ import java.io.PrintStream;
 import java.util.*;
 
 /**
- * Produces an input file to Beagle imputation engine, listing genotype likelihoods for each sample in input variant file
+ *  Converts the input VCF into a format accepted by the Beagle imputation/analysis program.
+ * <p>
+ *
+ * <h2>Input</h2>
+ * <p>
+ * A VCF with variants to convert to Beagle format
+ * </p>
+ *
+ * <h2>Outputs</h2>
+ * <p>
+ * A single text file which can be fed to Beagle
+ * </p>
+ * <p>
+ * Optional: A file with a list of markers
+ * </p>
+  *
+ * <h2>Examples</h2>
+ * <pre>
+ *     java -Xmx2g -jar dist/GenomeAnalysisTK.jar -L 20 \
+ *      -R reffile.fasta -T ProduceBeagleInput \
+ *      -V path_to_input_vcf/inputvcf.vcf -o path_to_beagle_output/beagle_output
+ * </pre>
+ *
  */
+
 public class ProduceBeagleInputWalker extends RodWalker<Integer, Integer> {
 
     @ArgumentCollection protected StandardVariantContextInputArgumentCollection variantCollection = new StandardVariantContextInputArgumentCollection();
 
-    @Input(fullName="validation", shortName = "validation", doc="Input VCF file", required=false)
+    @Hidden
+    @Input(fullName="validation", shortName = "validation", doc="Validation VCF file", required=false)
     public RodBinding<VariantContext> validation;
 
+
     @Output(doc="File to which BEAGLE input should be written",required=true)
     protected PrintStream  beagleWriter = null;
 
+    @Hidden
      @Output(doc="File to which BEAGLE markers should be written", shortName="markers", fullName = "markers", required = false)
     protected PrintStream  markers = null;
     int markerCounter = 1;
@@ -73,14 +99,19 @@ public class ProduceBeagleInputWalker extends RodWalker<Integer, Integer> {
     @Argument(doc="VQSqual key", shortName = "vqskey", required=false)
     protected String VQSLOD_KEY = "VQSqual";
 
+    @Hidden
      @Argument(fullName = "inserted_nocall_rate", shortName = "nc_rate", doc = "Rate (0-1) at which genotype no-calls will be randomly inserted, for testing", required = false)
     public double insertedNoCallRate  = 0;
+    @Hidden
      @Argument(fullName = "validation_genotype_ptrue", shortName = "valp", doc = "Flat probability to assign to validation genotypes. Will override GL field.", required = false)
     public double validationPrior = -1.0;
+    @Hidden
      @Argument(fullName = "validation_bootstrap", shortName = "bs", doc = "Proportion of records to be used in bootstrap set", required = false)
     public double bootstrap = 0.0;
+    @Hidden
      @Argument(fullName = "bootstrap_vcf",shortName = "bvcf", doc = "Output a VCF with the records used for bootstrapping filtered out", required = false)
     VCFWriter bootstrapVCFOutput = null;
+
     @Argument(fullName = "checkIsMaleOnChrX", shortName = "checkIsMaleOnChrX", doc = "Set to true when Beagle-ing chrX and want to ensure male samples don't have heterozygous calls.", required = false)
     public boolean CHECK_IS_MALE_ON_CHR_X = false;
 

public/java/src/org/broadinstitute/sting/utils/text/ListFileUtils.java

@@ -160,8 +160,8 @@ public class ListFileUtils {
                                 rodBinding.getName(), rodBinding.getTribbleType(), builderForValidation.userFriendlyListOfAvailableFeatures()));
             if ( ! rodBinding.getType().isAssignableFrom(descriptor.getFeatureClass()) )
                 throw new UserException.BadArgumentValue(rodBinding.getName(),
-                        String.format("Field %s expected type %s, but the type of the input file provided on the command line was %s. Please make sure that you have provided the correct file type and/or that you are not binding your rod to a name matching one of the available types.",
+                        String.format("Field %s expected type %s, but the type of the input file provided on the command line was %s producing %s. Please make sure that you have provided the correct file type and/or that you are not binding your rod to a name matching one of the available types.",
-                                rodBinding.getName(), rodBinding.getType(), descriptor.getName()));
+                                rodBinding.getName(), rodBinding.getType(), descriptor.getName(), descriptor.getFeatureClass()));
 
 
             rodBindings.add(triplet);

public/java/test/org/broadinstitute/sting/gatk/walkers/annotator/VariantAnnotatorIntegrationTest.java

@@ -133,7 +133,7 @@ public class VariantAnnotatorIntegrationTest extends WalkerTest {
             validationDataLocation + "1000G.exomes.vcf --snpEffFile  " + validationDataLocation +
             "snpEff_1.9.6_1000G.exomes.vcf_hg37.61.out -L 1:26,000,000-26,500,000",
             1,
-            Arrays.asList("c08648a078368c80530bff004b3157f1")
+            Arrays.asList("03eae1dab19a9358250890594bf53607")
         );
         executeTest("Testing SnpEff annotations", spec);
     }