analysis: remove HS metrics from alignment QC analysis, and generate …

…CLI, #TASK-2303, #TASK-2304
opencb · Nov 9, 2022 · cd4c96c · cd4c96c
1 parent ad9f2b8
commit cd4c96c
Show file tree

Hide file tree

Showing 80 changed files with 223 additions and 295 deletions.
diff --git a/...-analysis/src/main/java/org/opencb/opencga/analysis/alignment/qc/AlignmentQcAnalysis.java b/...-analysis/src/main/java/org/opencb/opencga/analysis/alignment/qc/AlignmentQcAnalysis.java
@@ -61,7 +61,6 @@ public class AlignmentQcAnalysis extends OpenCgaToolScopeStudy {
     private boolean runStats = true;
     private boolean runFlagStats = true;
     private boolean runFastqc = true;
-    private boolean runHsmetrics = true;
 
     @Override
     protected void check() throws Exception {
@@ -99,33 +98,10 @@ protected void check() throws Exception {
                         (!analysisParams.isOverwrite() && alignmentQc != null && alignmentQc.getFastQcMetrics() != null)) {
                     runFastqc = false;
                 }
-                if (skipValues.contains(AlignmentQcParams.HS_METRICS_SKIP_VALUE)
-                        ||
-                        (!analysisParams.isOverwrite() && alignmentQc != null && alignmentQc.getHsMetrics() != null)
-                        ||
-                        StringUtils.isEmpty(analysisParams.getBedFile())
-                        ||
-                        StringUtils.isEmpty(analysisParams.getDictFile())) {
-                    runHsmetrics = false;
-                }
             }
         } catch (CatalogException e) {
             throw new ToolException("Error accessing to the BAM file '" + analysisParams.getBamFile() + "'", e);
         }
-
-        if (runHsmetrics) {
-            try {
-                AnalysisUtils.getCatalogFile(analysisParams.getBedFile(), study, catalogManager.getFileManager(), token);
-            } catch (CatalogException e) {
-                throw new ToolException("Error accessing to the BED file '" + analysisParams.getBedFile() + "'", e);
-            }
-
-            try {
-                AnalysisUtils.getCatalogFile(analysisParams.getDictFile(), study, catalogManager.getFileManager(), token);
-            } catch (CatalogException e) {
-                throw new ToolException("Error accessing to the dictionary file '" + analysisParams.getDictFile() + "'", e);
-            }
-        }
     }
 
     @Override
@@ -136,7 +112,6 @@ protected void run() throws ToolException {
             String statsJobId = null;
             String flagStatsJobId = null;
             String fastQcMetricsJobId = null;
-            String hsMetricsJobId = null;
 
             try {
                 if (runFlagStats) {
@@ -189,29 +164,10 @@ protected void run() throws ToolException {
                 addWarning("Error launching job for Alignment FastQC Metrics Analysis: " + e.getMessage());
             }
 
-            try {
-                if (runHsmetrics) {
-                    // HS metrics
-                    params = new AlignmentHsMetricsParams(analysisParams.getBamFile(), analysisParams.getBedFile(),
-                            analysisParams.getDictFile(), null).toParams(new ObjectMap(ParamConstants.STUDY_PARAM, study));
-
-                    OpenCGAResult<Job> hsMetricsJobResult = catalogManager.getJobManager()
-                            .submit(study, AlignmentHsMetricsAnalysis.ID, Enums.Priority.MEDIUM, params, null,
-                                    "Job generated by " + getId() + " - " + getJobId(), Collections.emptyList(), Collections.emptyList(),
-                                    token);
-                    hsMetricsJobId = hsMetricsJobResult.first().getId();
-                    addEvent(Event.Type.INFO, "Submit job " + hsMetricsJobId + " to compute HS metrics (" + AlignmentHsMetricsAnalysis.ID
-                            + ")");
-                }
-            } catch (CatalogException e) {
-                addWarning("Error launching job for Alignment HS Metrics Analysis: " + e.getMessage());
-            }
-
             // Wait for those jobs before saving QC
             SamtoolsFlagstats samtoolsFlagstats = null;
             SamtoolsStats samtoolsStats = null;
             FastQcMetrics fastQcMetrics = null;
-            HsMetrics hsMetrics = null;
 
             if (flagStatsJobId != null) {
                 try {
@@ -249,17 +205,6 @@ protected void run() throws ToolException {
                     addWarning("Error waiting for job '" + fastQcMetricsJobId + "' (Alignment FastQC Metrics Analysis): " + e.getMessage());
                 }
             }
-            if (hsMetricsJobId != null) {
-                try {
-                    if (waitFor(hsMetricsJobId)) {
-                        Job job = getJob(hsMetricsJobId);
-                        logger.info("Alignment HS Metrics Analysis, job.outDir = " + job.getOutDir());
-                        hsMetrics = AlignmentHsMetricsAnalysis.parseResults(Paths.get(job.getOutDir().getUri().getPath()));
-                    }
-                } catch (Exception e) {
-                    addWarning("Error waiting for job '" + hsMetricsJobId + "' (Alignment FastQC Metrics Analysis): " + e.getMessage());
-                }
-            }
 
             // Update quality control for the catalog file
             catalogBamFile = AnalysisUtils.getCatalogFile(analysisParams.getBamFile(), study, catalogManager.getFileManager(), token);
@@ -284,10 +229,6 @@ protected void run() throws ToolException {
                 qc.getAlignment().setFastQcMetrics(fastQcMetrics);
                 saveQc = true;
             }
-            if (hsMetrics != null) {
-                qc.getAlignment().setHsMetrics(hsMetrics);
-                saveQc = true;
-            }
 
             if (saveQc) {
                 catalogManager.getFileManager().update(getStudy(), catalogBamFile.getId(), new FileUpdateParams().setQualityControl(qc),
@@ -324,7 +265,7 @@ private boolean waitFor(String jobId) throws ToolException {
         return status.equals(Enums.ExecutionStatus.DONE) ? true : false;
     }
 
-    private Job getJob(String jobId) throws ToolException {
+    private Job getJob(String jobId) {
         Job job = null;
         try {
             Query query = new Query("id", jobId);

diff --git a/...src/main/java/org/opencb/opencga/app/cli/internal/executors/AlignmentCommandExecutor.java b/...src/main/java/org/opencb/opencga/app/cli/internal/executors/AlignmentCommandExecutor.java
@@ -156,8 +156,6 @@ private void qcRun() throws ToolException {
 
         ObjectMap params = new AlignmentQcParams(
                 cliOptions.bamFile,
-                cliOptions.bedFile,
-                cliOptions.dictFile,
                 cliOptions.skip,
                 cliOptions.overwrite,
                 cliOptions.outdir

diff --git a/opencga-app/src/main/java/org/opencb/opencga/app/cli/main/OpenCgaCompleter.java b/opencga-app/src/main/java/org/opencb/opencga/app/cli/main/OpenCgaCompleter.java
@@ -1,5 +1,5 @@
 /*
-* Copyright 2015-2022-09-29 OpenCB
+* Copyright 2015-2022-11-09 OpenCB
 *
 * Licensed under the Apache License, Version 2.0 (the "License");
 * you may not use this file except in compliance with the License.

diff --git a/opencga-app/src/main/java/org/opencb/opencga/app/cli/main/OpencgaCliOptionsParser.java b/opencga-app/src/main/java/org/opencb/opencga/app/cli/main/OpencgaCliOptionsParser.java
@@ -1,5 +1,5 @@
 /*
-* Copyright 2015-2022-09-29 OpenCB
+* Copyright 2015-2022-11-09 OpenCB
 *
 * Licensed under the Apache License, Version 2.0 (the "License");
 * you may not use this file except in compliance with the License.

diff --git a/...main/java/org/opencb/opencga/app/cli/main/executors/AnalysisAlignmentCommandExecutor.java b/...main/java/org/opencb/opencga/app/cli/main/executors/AnalysisAlignmentCommandExecutor.java
@@ -481,8 +481,6 @@ private RestResponse<Job> runQc() throws Exception {
         } else {
             ObjectMap beanParams = new ObjectMap();
             putNestedIfNotEmpty(beanParams, "bamFile",commandOptions.bamFile, true);
-             putNestedIfNotEmpty(beanParams, "bedFile",commandOptions.bedFile, true);
-             putNestedIfNotEmpty(beanParams, "dictFile",commandOptions.dictFile, true);
              putNestedIfNotEmpty(beanParams, "skip",commandOptions.skip, true);
              putNestedIfNotNull(beanParams, "overwrite",commandOptions.overwrite, true);
              putNestedIfNotEmpty(beanParams, "outdir",commandOptions.outdir, true);

diff --git a/...rc/main/java/org/opencb/opencga/app/cli/main/options/AnalysisAlignmentCommandOptions.java b/...rc/main/java/org/opencb/opencga/app/cli/main/options/AnalysisAlignmentCommandOptions.java
@@ -421,7 +421,7 @@ public class RunPicardCommandOptions {
 
     }
 
-    @Parameters(commandNames = {"qc-run"}, commandDescription ="Compute quality control (QC) metrics for a given alignment file (including samtools stats, samtools flag stats, FastQC and HS metrics)")
+    @Parameters(commandNames = {"qc-run"}, commandDescription ="Compute quality control (QC) metrics for a given alignment file: samtools stats, samtools flag stats and FastQC metrics.")
     public class RunQcCommandOptions {
 
         @ParametersDelegate
@@ -448,22 +448,16 @@ public class RunQcCommandOptions {
         @Parameter(names = {"--job-tags"}, description = "Job tags", required = false, arity = 1)
         public String jobTags; 
 
-        @Parameter(names = {"--bam-file"}, description = "The body web service bamFile parameter", required = false, arity = 1)
+        @Parameter(names = {"--bam-file"}, description = "ID for the BAM file to process.", required = false, arity = 1)
         public String bamFile;
 
-        @Parameter(names = {"--bed-file"}, description = "The body web service bedFile parameter", required = false, arity = 1)
-        public String bedFile;
-
-        @Parameter(names = {"--dict-file"}, description = "The body web service dictFile parameter", required = false, arity = 1)
-        public String dictFile;
-
-        @Parameter(names = {"--skip"}, description = "The body web service skip parameter", required = false, arity = 1)
+        @Parameter(names = {"--skip"}, description = "To skip any alignment QC metrics use the following keywords (separated by commas): stats, flagstats, fastqc", required = false, arity = 1)
         public String skip;
 
-        @Parameter(names = {"--overwrite"}, description = "The body web service overwrite parameter", required = false, help = true, arity = 0)
+        @Parameter(names = {"--overwrite"}, description = "To overwrite the QC metrics already computed.", required = false, help = true, arity = 0)
         public boolean overwrite = false;
 
-        @Parameter(names = {"--outdir"}, description = "The body web service outdir parameter", required = false, arity = 1)
+        @Parameter(names = {"--outdir"}, description = "Output dir for the job.", required = false, arity = 1)
         public String outdir;
 
     }

diff --git a/...src/main/java/org/opencb/opencga/app/cli/main/options/AnalysisClinicalCommandOptions.java b/...src/main/java/org/opencb/opencga/app/cli/main/options/AnalysisClinicalCommandOptions.java
@@ -1613,7 +1613,7 @@ public class QueryVariantCommandOptions {
         @Parameter(names = {"--file-data"}, description = "Filter by file data (i.e. FILTER, QUAL and INFO columns from VCF file). [{file}:]{key}{op}{value}[,;]* . If no file is specified, will use all files from 'file' filter. e.g. AN>200 or file_1.vcf:AN>200;file_2.vcf:AN<10 . Many fields can be combined. e.g. file_1.vcf:AN>200;DB=true;file_2.vcf:AN<10,FILTER=PASS,LowDP", required = false, arity = 1)
         public String fileData; 
 
-        @Parameter(names = {"--sample"}, description = "Filter variants by sample genotype. This will automatically set 'includeSample' parameter when not provided. This filter accepts multiple 3 forms: 1) List of samples: Samples that contain the main variant. Accepts AND (;) and OR (,) operators.  e.g. HG0097,HG0098 . 2) List of samples with genotypes: {sample}:{gt1},{gt2}. Accepts AND (;) and OR (,) operators.  e.g. HG0097:0/0;HG0098:0/1,1/1 . Unphased genotypes (e.g. 0/1, 1/1) will also include phased genotypes (e.g. 0|1, 1|0, 1|1), but not vice versa. When filtering by multi-allelic genotypes, any secondary allele will match, regardless of its position e.g. 1/2 will match with genotypes 1/2, 1/3, 1/4, .... Genotype aliases accepted: HOM_REF, HOM_ALT, HET, HET_REF, HET_ALT, HET_MISS and MISS  e.g. HG0097:HOM_REF;HG0098:HET_REF,HOM_ALT . 3) Sample with segregation mode: {sample}:{segregation}. Only one sample accepted.Accepted segregation modes: [ autosomalDominant, autosomalRecessive, XLinkedDominant, XLinkedRecessive, YLinked, mitochondrial, deNovo, mendelianError, compoundHeterozygous ]. Value is case insensitive. e.g. HG0097:DeNovo Sample must have parents defined and indexed. ", required = false, arity = 1)
+        @Parameter(names = {"--sample"}, description = "Filter variants by sample genotype. This will automatically set 'includeSample' parameter when not provided. This filter accepts multiple 3 forms: 1) List of samples: Samples that contain the main variant. Accepts AND (;) and OR (,) operators.  e.g. HG0097,HG0098 . 2) List of samples with genotypes: {sample}:{gt1},{gt2}. Accepts AND (;) and OR (,) operators.  e.g. HG0097:0/0;HG0098:0/1,1/1 . Unphased genotypes (e.g. 0/1, 1/1) will also include phased genotypes (e.g. 0|1, 1|0, 1|1), but not vice versa. When filtering by multi-allelic genotypes, any secondary allele will match, regardless of its position e.g. 1/2 will match with genotypes 1/2, 1/3, 1/4, .... Genotype aliases accepted: HOM_REF, HOM_ALT, HET, HET_REF, HET_ALT, HET_MISS and MISS  e.g. HG0097:HOM_REF;HG0098:HET_REF,HOM_ALT . 3) Sample with segregation mode: {sample}:{segregation}. Only one sample accepted.Accepted segregation modes: [ autosomalDominant, autosomalRecessive, XLinkedDominant, XLinkedRecessive, YLinked, mitochondrial, deNovo, deNovoStrict, mendelianError, compoundHeterozygous ]. Value is case insensitive. e.g. HG0097:DeNovo Sample must have parents defined and indexed. ", required = false, arity = 1)
         public String sample; 
 
         @Parameter(names = {"--sample-data"}, description = "Filter by any SampleData field from samples. [{sample}:]{key}{op}{value}[,;]* . If no sample is specified, will use all samples from 'sample' or 'genotype' filter. e.g. DP>200 or HG0097:DP>200,HG0098:DP<10 . Many FORMAT fields can be combined. e.g. HG0097:DP>200;GT=1/1,0/1,HG0098:DP<10", required = false, arity = 1)
@@ -1655,7 +1655,7 @@ public class QueryVariantCommandOptions {
         @Parameter(names = {"--family-disorder"}, description = "Specify the disorder to use for the family segregation", required = false, arity = 1)
         public String familyDisorder; 
 
-        @Parameter(names = {"--family-segregation"}, description = "Filter by segregation mode from a given family. Accepted values: [ autosomalDominant, autosomalRecessive, XLinkedDominant, XLinkedRecessive, YLinked, mitochondrial, deNovo, mendelianError, compoundHeterozygous ]", required = false, arity = 1)
+        @Parameter(names = {"--family-segregation"}, description = "Filter by segregation mode from a given family. Accepted values: [ autosomalDominant, autosomalRecessive, XLinkedDominant, XLinkedRecessive, YLinked, mitochondrial, deNovo, deNovoStrict, mendelianError, compoundHeterozygous ]", required = false, arity = 1)
         public String familySegregation; 
 
         @Parameter(names = {"--family-members"}, description = "Sub set of the members of a given family", required = false, arity = 1)