repaired starfusion bug and hg38 exome somatic only bug

CCBR · Sep 21, 2018 · bf2950d · bf2950d
1 parent 5dfa182
commit bf2950d
Show file tree

Hide file tree

Showing 8 changed files with 50 additions and 18 deletions.
diff --git a/Rules/all-rnafusion.rl b/Rules/all-rnafusion.rl
@@ -21,7 +21,7 @@ if config['project']['annotation'] == "hg19":
            expand("fusioncatcher/{x}/final-list_candidate-fusion-genes.txt",x=samples),
            expand("starfusion/fusioninspector/{x}/{x}.fusion_predictions.final",x=samples),
            expand("fusioncatcher/fusioninspector/{x}/{x}.fusion_predictions.final",x=samples),
-           expand("starfusion/{x}/star-fusion.fusion_candidates.final",x=samples),
+           expand("starfusion/{x}/star-fusion.fusion_predictions.tsv",x=samples),
            expand("{name}.RnaSeqMetrics.txt",name=samples),
 #           "expression/RawCountFile_genes_filtered.txt",
 #           "expression/RawCountFile_junctions_filtered.txt",
@@ -62,7 +62,7 @@ elif config['project']['annotation'] == "hg38":
            expand("fusioncatcher/{x}/final-list_candidate-fusion-genes.txt",x=samples),
            expand("starfusion/fusioninspector/{x}/{x}.fusion_predictions.final",x=samples),
            expand("fusioncatcher/fusioninspector/{x}/{x}.fusion_predictions.final",x=samples),
-           expand("starfusion/{x}/star-fusion.fusion_candidates.final",x=samples),
+           expand("starfusion/{x}/star-fusion.fusion_predictions.tsv",x=samples),
            expand("{name}.RnaSeqMetrics.txt",name=samples),
 #           "expression/RawCountFile_genes_filtered.txt",
 #           "expression/RawCountFile_junctions_filtered.txt",
@@ -99,7 +99,7 @@ elif config['project']['annotation'] == "mm10":
            expand("fusioncatcher/{x}/final-list_candidate-fusion-genes.txt",x=samples),
            expand("starfusion/fusioninspector/{x}/{x}.fusion_predictions.final",x=samples),
            expand("fusioncatcher/fusioninspector/{x}/{x}.fusion_predictions.final",x=samples),
-           expand("starfusion/{x}/star-fusion.fusion_candidates.final",x=samples),
+           expand("starfusion/{x}/star-fusion.fusion_predictions.tsv",x=samples),
            expand("{name}.RnaSeqMetrics.txt",name=samples),
 #           "expression/RawCountFile_genes_filtered.txt",
 #           "expression/RawCountFile_junctions_filtered.txt",

diff --git a/Rules/all-wgslow.rl b/Rules/all-wgslow.rl
@@ -16,6 +16,16 @@ if config['project']['annotation'] == "hg19":
            expand("{s}"+".g.vcf",s=samples),
            expand("canvas_out/{s}/CNV.vcf.gz", s=samples),
            "admixture_out/admixture_table.tsv",
+           expand("{s}.recal.bam",s=samples),
+           expand("QC/{s}.R1_fastqc.html",s=samples),
+#          expand("QC/{s}.R1."+config['project']['filetype']+"_filtered",s=samples),
+#          expand("{s}.R1.trimmed.fastq.bz2",s=samples),
+#          expand("QC/{s}.qualimapReport",s=samples),
+           expand("QC/{s}.{r}.trimmed_fastqc.html",s=samples,r=['R1','R1']),
+           expand("QC/{s}.{r}.trimmed_screen.txt",s=samples,r=['R1','R1']),
+           expand("QC/{s}.{r}.trimmed_screen.png",s=samples,r=['R1','R1']),
+           expand("QC/{s}_run_trimmomatic.err",s=samples),
+           expand("QC/{s}.qualimapReport/genome_results.txt",s=samples),
 
     output: 
     params: rname="final"
@@ -41,7 +51,17 @@ elif config['project']['annotation'] == "hg38":
            config['project']['workpath']+"/svaba_out/svaba.log",
            expand("{s}"+".g.vcf",s=samples),
            expand("canvas_out/{s}/CNV.vcf.gz", s=samples),
-           "admixture_out/admixture_table.tsv"
+           "admixture_out/admixture_table.tsv",
+           expand("{s}.recal.bam",s=samples),
+           expand("QC/{s}.R1_fastqc.html",s=samples),
+#          expand("QC/{s}.R1."+config['project']['filetype']+"_filtered",s=samples),
+#          expand("{s}.R1.trimmed.fastq.bz2",s=samples),
+#          expand("QC/{s}.qualimapReport",s=samples),
+           expand("QC/{s}.{r}.trimmed_fastqc.html",s=samples,r=['R1','R1']),
+           expand("QC/{s}.{r}.trimmed_screen.txt",s=samples,r=['R1','R1']),
+           expand("QC/{s}.{r}.trimmed_screen.png",s=samples,r=['R1','R1']),
+           expand("QC/{s}_run_trimmomatic.err",s=samples),
+           expand("QC/{s}.qualimapReport/genome_results.txt",s=samples),
 
     output: 
     params: rname="final"
@@ -66,7 +86,17 @@ elif config['project']['annotation'] == "mm10":
            config['project']['workpath']+"/delly_out/insertions.bcf",
            config['project']['workpath']+"/svaba_out/svaba.log",
            expand("canvas_out/{s}/CNV.vcf.gz", s=samples),
-           "admixture_out/admixture_table.tsv"
+           "admixture_out/admixture_table.tsv",
+           expand("{s}.recal.bam",s=samples),
+           expand("QC/{s}.R1_fastqc.html",s=samples),
+#          expand("QC/{s}.R1."+config['project']['filetype']+"_filtered",s=samples),
+#          expand("{s}.R1.trimmed.fastq.bz2",s=samples),
+#          expand("QC/{s}.qualimapReport",s=samples),
+           expand("QC/{s}.{r}.trimmed_fastqc.html",s=samples,r=['R1','R1']),
+           expand("QC/{s}.{r}.trimmed_screen.txt",s=samples,r=['R1','R1']),
+           expand("QC/{s}.{r}.trimmed_screen.png",s=samples,r=['R1','R1']),
+           expand("QC/{s}_run_trimmomatic.err",s=samples),
+           expand("QC/{s}.qualimapReport/genome_results.txt",s=samples),
     output: 
     params: rname="final"
     shell:  """

diff --git a/Rules/fusioninsp_starfus.rl b/Rules/fusioninsp_starfus.rl
@@ -1,5 +1,5 @@
 rule rule fusioninsp_starfus:
-     input:  fusions="starfusion/{x}/star-fusion.fusion_candidates.final",file1="{x}.R1.trimmed.fastq.gz",file2="{x}.R2.trimmed.fastq.gz"
+     input:  fusions="starfusion/{x}/star-fusion.fusion_predictions.tsv",file1="{x}.R1.trimmed.fastq.gz",file2="{x}.R2.trimmed.fastq.gz"
      output: "starfusion/fusioninspector/{x}/{x}.fusion_predictions.final"
      params: rname='fusioninsp',sample="{x}",starlib=config['references'][pfamily]['STARFUSIONLIB']
      shell: "module load fusioninspector/1.1.0; module load python/2.7; mkdir -p starfusion/fusioninspector/{params.sample}; FusionInspector --fusions {input.fusions} --genome_lib {params.starlib} --left_fq {input.file1} --right_fq {input.file2} --out_dir starfusion/fusioninspector/{params.sample} --out_prefix {params.sample} --prep_for_IGV"
diff --git a/Rules/qualimap.rl b/Rules/qualimap.rl
@@ -1,5 +1,5 @@
 rule qualimap:
-    input:  bam="{x}.dedup.bam",
+    input:  bam="{x}.recal.bam",
             bed=config['project']['workpath']+"/exome_targets.bed"
     output: dir="QC/{x}.qualimapReport",txt="QC/{x}.qualimapReport/genome_results.txt"
     threads: 8

diff --git a/Rules/qualimap_wgs.rl b/Rules/qualimap_wgs.rl
@@ -1,5 +1,5 @@
 rule qualimap_wgs:
-    input:  "{x}.dedup.bam"
+    input:  "{x}.recal.bam"
     output: "QC/{x}.qualimapReport","QC/{x}.qualimapReport/genome_results.txt"
     threads: 8
     params: qualimap=config['bin'][pfamily]['QUALIMAP'],organism=config['references'][pfamily]['ORGANISM'],rname="pl:wgs_qualimap"

diff --git a/Rules/starfusion.rl b/Rules/starfusion.rl
@@ -1,5 +1,5 @@
 rule starfusion:
      input:  file1="{x}.R1.trimmed.fastq.gz",file2="{x}.R2.trimmed.fastq.gz"
-     output: "starfusion/{x}/star-fusion.fusion_candidates.final"
+     output: "starfusion/{x}/star-fusion.fusion_predictions.tsv"
      params: rname='starfusion',sample="{x}",starlib=config['references'][pfamily]['STARFUSIONLIB']
      shell: "module load STAR-Fusion/1.3.2; module load STAR/2.5.4a; mkdir -p starfusion/{params.sample}; STAR-Fusion --genome_lib_dir {params.starlib} --left_fq {input.file1} --right_fq {input.file2} --output_dir starfusion/{params.sample} --CPU 32"
diff --git a/hg38.json b/hg38.json
@@ -8,7 +8,7 @@
     "CNVKITACCESS": "/data/CCBR_Pipeliner/db/PipeDB/lib/access-5k-mappable.grch38.bed",
     "INDELSITES": "/fdb/GATK_resource_bundle/hg38bundle/Mills_and_1000G_gold_standard.indels.hg38.vcf.gz",
 #    "NOVOINDEX": "/data/CCBR_Pipeliner/db/PipeDB/lib/human_g1k_v37_iupac.nix", 
-#    "REFFLAT": "/data/CCBR_Pipeliner/db/PipeDB/lib/reformatS04380219_qualimap.bed",
+    "REFFLAT": "/data/CCBR_Pipeliner/db/PipeDB/lib/reformatS04380219_qualimap.bed",
     "KNOWNANCESTRY": "/data/CCBR_Pipeliner/db/PipeDB/lib/1k_genomes_phase3_autosomes.hg38.vcf.gz",
     "ADMIXTUREKEY": "/data/CCBR_Pipeliner/db/PipeDB/lib/1k_genomes_superpop_key.txt",
     "ADMIXTUREREFS": "5",
@@ -79,7 +79,7 @@
     "CNVKITANNO": "/data/CCBR_Pipeliner/db/PipeDB/lib/hg38_refFlat.txt",
     "INDELSITES": "/fdb/GATK_resource_bundle/hg38bundle/Mills_and_1000G_gold_standard.indels.hg38.vcf.gz",
 #    "NOVOINDEX": "/data/CCBR_Pipeliner/db/PipeDB/lib/human_g1k_v37_iupac.nix", 
-#    "REFFLAT": "/data/CCBR_Pipeliner/db/PipeDB/lib/reformatS04380219_qualimap.bed",
+    "REFFLAT": "/data/CCBR_Pipeliner/db/PipeDB/lib/reformatS04380219_qualimap.bed",
     "KNOWNANCESTRY": "/data/CCBR_Pipeliner/db/PipeDB/lib/1k_genomes_phase3_autosomes.hg38.vcf.gz",
     "ADMIXTUREKEY": "/data/CCBR_Pipeliner/db/PipeDB/lib/1k_genomes_superpop_key.txt",
     "ADMIXTUREREFS": "5",
@@ -130,6 +130,7 @@
     "VEPFILTERVCF": "/data/CCBR_Pipeliner/db/PipeDB/lib/ExAC_nonTCGA.r0.3.1.sites.vep.GRCh38.vcf.gz",
     "VEPGENOME": "/fdb/VEP/88/cache/GRCh38.fa",
     "CANVASGENOME": "/data/CCBR_Pipeliner/db/PipeDB/lib/canvas_refs/GRCh38/WholeGenomeFasta",
+    "CANVASPLOIDY": "/data/CCBR_Pipeliner/db/PipeDB/lib/canvas_knownPloidy_hg19.vcf",
     "CANVASKMER": "/data/CCBR_Pipeliner/db/PipeDB/lib/canvas_refs/GRCh38/kmer.fa",
     "CANVASFILTER": "/data/CCBR_Pipeliner/db/PipeDB/lib/canvas_refs/GRCh38/filter13.bed",
     "CNVKITANNO": "/data/CCBR_Pipeliner/db/PipeDB/lib/hg38_refFlat.txt",

diff --git a/rules.json b/rules.json
@@ -1,6 +1,7 @@
  {
     "rules": {
         "avia": ["wgslow","exomeseq-germline","exomeseq-germline-recal","exomeseq-germline-partial","rnaseqvargerm"],
+        "index_bams": ["wgslow","exomeseq-germline","wgs-somatic-tumoronly","exomeseq-somatic-tumoronly","wgs-somatic","exomeseq-somatic"],
         "rnaseqforfusions": ["rnaseqfusion"],
         "rnafusioncleanup": ["none"],
         "vcf2maf": ["exomeseq-somatic","wgs-somatic"],
@@ -31,11 +32,11 @@
         "gatk_select_variants_rnaseqvar": ["rnaseqvargerm"],
         "gatk.merge.chrom": ["exomeseq-germline","exomeseq-germline-recal","exomeseq-germline-partial","wgslow","rnaseqvargerm"],
         "ngsqc": ["initialqc","initialqcgenomeseq","rnaseqvargerm"],
-        "fastqc.fastq": ["initialqc","initialqcgenomeseq"],
+        "fastqc.fastq": ["initialqc","initialqcgenomeseq","wgslow","exomeseq-germline","wgs-somatic-tumoronly","exomeseq-somatic-tumoronly","wgs-somatic","exomeseq-somatic"],
         "fastqc.rnaseqvar": ["rnaseqvargerm","rnaseqfusion"],
-        "bamstats": ["initialqc"],
+        "bamstats": ["initialqc","exomeseq-germline","exomeseq-somatic-tumoronly","exomeseq-somatic"],
         "make_target_files": ["initialqc","exomeseq-somatic","exomeseq-somatic-tumoronly","exomeseq-germline"],
-        "fastq_screen": ["initialqc","initialqcgenomeseq","rnaseqvargerm","rnaseqfusion"],
+        "fastq_screen": ["initialqc","initialqcgenomeseq","rnaseqvargerm","rnaseqfusion","wgslow","exomeseq-germline","wgs-somatic-tumoronly","exomeseq-somatic-tumoronly","wgs-somatic","exomeseq-somatic"],
         "multiqc": ["initialqc"],
         "multiqc_genomeseq": ["initialqcgenomeseq"],
         "alview": ["exomeseq-somatic"],
@@ -62,13 +63,13 @@
         "cnvkit_somatic_tumoronly": ["exomeseq-somatic-tumoronly"],
         "cnvkit_summary_tumoronly": ["exomeseq-somatic-tumoronly"],
         "delly_somatic": ["exomeseq-somatic","wgs-somatic"],
-        "fastqc.trimmed": ["initialqc","initialqcgenomeseq","rnaseqvargerm","rnaseqfusion"],
-        "trimmomatic": ["initialqc","initialqcgenomeseq","rnaseqvargerm","rnaseqfusion"],
+        "fastqc.trimmed": ["initialqc","initialqcgenomeseq","rnaseqvargerm","rnaseqfusion","wgslow","exomeseq-germline","wgs-somatic-tumoronly","exomeseq-somatic-tumoronly","wgs-somatic","exomeseq-somatic"],
+        "trimmomatic": ["initialqc","initialqcgenomeseq","rnaseqvargerm","rnaseqfusion","wgslow","exomeseq-germline","wgs-somatic-tumoronly","exomeseq-somatic-tumoronly","wgs-somatic","exomeseq-somatic"],
         "samtools.flagstats": ["none"],
         "samtools.flagstats.dedup": ["none"],
         "qualimap_rnaseqvar": ["rnaseqvargerm"],
-        "qualimap_wgs": ["initialqcgenomeseq"],
-        "qualimap": ["initialqc"],
+        "qualimap_wgs": ["initialqcgenomeseq","wgslow","wgs-somatic-tumoronly","wgs-somatic"],
+        "qualimap": ["initialqc","exomeseq-germline","exomeseq-somatic-tumoronly","exomeseq-somatic"],
         "bwa.mem": ["initialqc","initialqcgenomeseq"],
         "novocraft.novoalign": ["none"],
         "bwa.index.ref": ["none"],