DSL2 - Genotyping Input Updates

conf/test_krakenuniq.config

@@ -32,5 +32,5 @@ params {
     // Metagenomics
     run_metagenomics                = true
     metagenomics_profiling_tool     = 'krakenuniq'
-    metagenomics_profiling_database = params.pipelines_testdata_base_path + 'eager/databases/krakenuniq/testdb-krakenuniq.tar.gz'
+    metagenomics_profiling_database = params.pipelines_testdata_base_path + 'eager/databases/krakenuniq/eager3-mammoth-minimal.tar.gz'
 }

conf/test_metaphlan.config

@@ -24,7 +24,7 @@ params {
     config_profile_description      = 'Minimal test dataset to check the metagenomics metaphlan4 pipeline function'
 
     // Input data
-    input                           = params.pipelines_testdata_base_path + 'eager/testdata/Mammoth/samplesheet_v3.tsv'
+    input                           = params.pipelines_testdata_base_path + 'eager/testdata/Mammoth/samplesheet_v3_TOYMETAPHLAN.csv'
 
     // Genome references
     fasta                           = params.pipelines_testdata_base_path + 'eager/reference/Mammoth/Mammoth_MT_Krause.fasta'

docs/development/manual_tests.md

@@ -721,7 +721,7 @@ HOP001 ERR8958750 0 4 paired double half /workspace/eager/testing/test_data/ERR8
 HOP001 ERR8958751 0 2 paired double half /workspace/eager/testing/test_data/ERR8958751_1.fastq.gz_reduced.fastq.gz /workspace/eager/testing/test_data/ERR8958751_2.fastq.gz_reduced.fastq.gz NA NA
 HOP001 ERR8958752 0 2 paired double half /workspace/eager/testing/test_data/ERR8958752_1.fastq.gz_reduced.fastq.gz /workspace/eager/testing/test_data/ERR8958752_2.fastq.gz_reduced.fastq.gz NA NA
 HOP001 ERR8958753 0 2 paired double half /workspace/eager/testing/test_data/ERR8958753_1.fastq.gz_reduced.fastq.gz /workspace/eager/testing/test_data/ERR8958753_2.fastq.gz_reduced.fastq.gz NA NA
-HOP001 ERR8958754 0 2 paired double none /workspace/eager/testing/test_data/ERR8958754_1.fastq.gz_reduced.fastq.gz /workspace/eager/testing/test_data/ERR8958754_2.fastq.gz_reduced.fastq.gz NA NA" | sed 's/ /\t/g' > test.tsv
+HOP001 ERR8958754 0 2 paired double none /workspace/eager/testing/test_data/ERR8958754_1.fastq.gz_reduced.fastq.gz /workspace/eager/testing/test_data/ERR8958754_2.fastq.gz_reduced.fastq.gz NA NA" | sed 's/NA/ /g' | sed 's/ /\t/g'  > test.tsv
 
 nextflow run ../main.nf -profile docker \
   --input test.tsv \
@@ -738,6 +738,16 @@ nextflow run ../main.nf -profile docker \
   --metagenomics_malt_group_size 3
 ```
 
+# kraken2
+
+nextflow run main.nf -profile docker \
+ --input testing/test.tsv \
+ --outdir ./out \
+ --run_metagenomics \
+ --metagenomics_profiling_tool kraken2 \
+ --metagenomics_profiling_database /workspace/eager/testing/eager_test.tar.gz
+--preprocessing_skippairmerging
+
 ## Mapping statistics
 
 ### ENDOSPY
@@ -981,7 +991,7 @@ nextflow run main.nf -profile test,docker --outdir ./results -w work/ -resume --
 ## Gatk UG on trimmed reads. Skip bcftools stats.
 ## Expect: One VCF + .tbi index per sample/reference combination, based on the trimmed bams (this actually shows on the IndelRealigner step and not the UG step).  No IR directory. No bcftools_stats file per VCF.
 ## Checked that the input bam for the UG jobs indeed had trimmed reads. (The full UDG sample has untrimmed bams.)
-nextflow run main.nf -profile test,docker --outdir ./results -w work/ -resume --run_genotyping --genotyping_tool 'ug' --genotyping_source 'trimmed' -ansi-log false -dump-channels --skip_bcftools_stats \
+nextflow run main.nf -profile test,docker --outdir ./results -w work/ -resume --run_genotyping --genotyping_tool 'ug' --genotyping_source 'trimmed' -ansi-log false -dump-channels --genotyping_skip_bcftools_stats \
   --run_trim_bam \
   --damage_manipulation_bamutils_trim_double_stranded_none_udg_left 5 \
   --damage_manipulation_bamutils_trim_double_stranded_none_udg_right 7 \
@@ -1020,7 +1030,7 @@ nextflow run main.nf -profile test,docker --outdir ./results -w work/ -resume --
 ```bash
 ## Attempt to run Gatk HC on trimmed reads, without activating trimming.
 ## Expect: FAILURE. Cannot set genotyping source to ;trimmed' without trimming.
-nextflow run main.nf -profile test,docker --outdir ./results -w work/ -resume --run_genotyping --genotyping_tool 'hc' --genotyping_source 'trimmed' -ansi-log false -dump-channels --skip_bcftools_stats \
+nextflow run main.nf -profile test,docker --outdir ./results -w work/ -resume --run_genotyping --genotyping_tool 'hc' --genotyping_source 'trimmed' -ansi-log false -dump-channels --genotyping_skip_bcftools_stats \
   --genotyping_gatk_hc_emitrefconf 'BP_RESOLUTION' \
   --genotyping_gatk_hc_out_mode 'EMIT_ALL_ACTIVE_SITES'
 ```
@@ -1029,7 +1039,7 @@ nextflow run main.nf -profile test,docker --outdir ./results -w work/ -resume --
 ## Gatk HC on trimmed reads, with different out mode and emit confidence. Skip bcftools stats.
 ## Expect: One VCF + .tbi index per sample/reference combination.
 ## Checked .command.sh for correct args.
-nextflow run main.nf -profile test,docker --outdir ./results -w work/ -resume --run_genotyping --genotyping_tool 'hc' --genotyping_source 'trimmed' --run_trim_bam -ansi-log false -dump-channels --skip_bcftools_stats \
+nextflow run main.nf -profile test,docker --outdir ./results -w work/ -resume --run_genotyping --genotyping_tool 'hc' --genotyping_source 'trimmed' --run_trim_bam -ansi-log false -dump-channels --genotyping_skip_bcftools_stats \
   --genotyping_gatk_hc_emitrefconf 'BP_RESOLUTION' \
   --genotyping_gatk_hc_out_mode 'EMIT_ALL_ACTIVE_SITES'
 ```
@@ -1053,7 +1063,7 @@ nextflow run main.nf -profile test,docker --outdir ./results -w work/ -resume --
 ## Freebayes on trimmed reads. Different options, and skip bcftools stats.
 ## Expect: One VCF + .tbi index per sample/reference combination.
 ## Checked .command.sh for correct args.
-nextflow run main.nf -profile test,docker --outdir ./results -w work/ -resume --run_genotyping --genotyping_tool 'freebayes' --genotyping_source 'trimmed' -ansi-log false -dump-channels --skip_bcftools_stats \
+nextflow run main.nf -profile test,docker --outdir ./results -w work/ -resume --run_genotyping --genotyping_tool 'freebayes' --genotyping_source 'trimmed' -ansi-log false -dump-channels --genotyping_skip_bcftools_stats \
   --run_trim_bam \
   --genotyping_freebayes_skip_coverage 10 \
   --genotyping_freebayes_min_alternate_count 2 \

modules.json

@@ -187,7 +187,7 @@
                     },
                     "krakenuniq/preloadedkrakenuniq": {
                         "branch": "master",
-                        "git_sha": "a6eb17f65b3ee5761c25c075a6166c9f76733cee",
+                        "git_sha": "666652151335353eef2fcd58880bcef5bc2928e1",
                         "installed_by": ["modules"]
                     },
                     "malt/run": {
@@ -225,6 +225,11 @@
                         "git_sha": "f0719ae309075ae4a291533883847c3f7c441dad",
                         "installed_by": ["modules"]
                     },
+                    "picard/addorreplacereadgroups": {
+                        "branch": "master",
+                        "git_sha": "81880787133db07d9b4c1febd152c090eb8325dc",
+                        "installed_by": ["modules"]
+                    },
                     "picard/createsequencedictionary": {
                         "branch": "master",
                         "git_sha": "20b0918591d4ba20047d7e13e5094bcceba81447",