adds protected tags - #9

workflow/rules/aggregate_results.smk

@@ -10,10 +10,10 @@ rule multiqc_by_sample_initial_pass:
         multiqc_config="configs/multiqc_config.yaml",
         rename_config=PROJECT_PATH / "{sample}" / "qc" / "multiqc_initial_pass" / "multiqc_sample_rename_config"  / "{sample}_rename_config.tsv",
     output:
-        OUT_DIR / "{sample}" / "qc" / "multiqc_initial_pass" / "{sample}_multiqc.html",
-        OUT_DIR / "{sample}" / "qc" / "multiqc_initial_pass" / "{sample}_multiqc_data" / "multiqc_general_stats.txt",
-        OUT_DIR / "{sample}" / "qc" / "multiqc_initial_pass" / "{sample}_multiqc_data" / "multiqc_fastqc_trimmed.txt",
-        OUT_DIR / "{sample}" / "qc" / "multiqc_initial_pass" / "{sample}_multiqc_data" / "multiqc_fastq_screen.txt",
+        protected(OUT_DIR / "{sample}" / "qc" / "multiqc_initial_pass" / "{sample}_multiqc.html"),
+        protected(OUT_DIR / "{sample}" / "qc" / "multiqc_initial_pass" / "{sample}_multiqc_data" / "multiqc_general_stats.txt"),
+        protected(OUT_DIR / "{sample}" / "qc" / "multiqc_initial_pass" / "{sample}_multiqc_data" / "multiqc_fastqc_trimmed.txt"),
+        protected(OUT_DIR / "{sample}" / "qc" / "multiqc_initial_pass" / "{sample}_multiqc_data" / "multiqc_fastq_screen.txt"),
     # WARNING: don't put this rule in a group, bad things will happen. see issue #23 in gitlab (small var caller pipeline repo)
     message:
         "Aggregates QC results using multiqc. First pass. Output will be used for the internal QC checkup. Sample: {wildcards.sample}"
@@ -33,10 +33,10 @@ rule qc_checkup:
         fastq_screen=OUT_DIR / "{sample}" / "qc" / "multiqc_initial_pass" / "{sample}_multiqc_data" / "multiqc_fastq_screen.txt",
         qualimap=OUT_DIR / "{sample}" / "qc" / "qualimap" / "{sample}" / "genome_results.txt",
     output:
-        expand(
+        protected(expand(
             OUT_DIR / "{{sample}}" / "qc" / "qc_checkup" / "qc_checkup_{suffix}.yaml",
             suffix=["overall_summary", "fastqc", "fastq_screen", "qualimap_overall", "qualimap_chromosome_stats"],
-        ),
+        )),
     # WARNING: don't put this rule in a group, bad things will happen. see issue #23 in gitlab
     message:
         "Runs QC checkup on various QC tool output, based on custom defined QC thresholds. "
@@ -72,8 +72,8 @@ rule multiqc_by_sample_final_pass:
         rename_config=PROJECT_PATH / "{sample}" / "qc" / "multiqc_initial_pass" / "multiqc_sample_rename_config"  / "{sample}_rename_config.tsv",
         qc_config="configs/qc_checkup/qc_checkup_config.yaml",
     output:
-        OUT_DIR / "{sample}" / "qc" / "multiqc_final_pass" / "{sample}_multiqc.html",
-        OUT_DIR / "{sample}" / "qc" / "multiqc_final_pass" / "{sample}_multiqc_data" / "multiqc_general_stats.txt",
+        protected(OUT_DIR / "{sample}" / "qc" / "multiqc_final_pass" / "{sample}_multiqc.html"),
+        protected(OUT_DIR / "{sample}" / "qc" / "multiqc_final_pass" / "{sample}_multiqc_data" / "multiqc_general_stats.txt"),
     # WARNING: don't put this rule in a group, bad things will happen. see issue #23 in gitlab
     message:
         "Aggregates QC results using multiqc. Final pass, where QC checkup results are also aggregated"
@@ -94,7 +94,7 @@ rule aggregate_sample_rename_configs:
     input:
         expand(PROJECT_PATH / "{sample}" / "qc" / "multiqc_initial_pass" / "multiqc_sample_rename_config"  / "{sample}_rename_config.tsv", sample=SAMPLES),
     output:
-        OUT_DIR / "project_level_qc" / "multiqc" / "aggregated_rename_configs.tsv",
+        protected(OUT_DIR / "project_level_qc" / "multiqc" / "aggregated_rename_configs.tsv"),
     message:
         "Aggregate all sample rename-config files"
     run:
@@ -125,7 +125,7 @@ rule multiqc_aggregation_all_samples:
         multiqc_config="configs/multiqc_config.yaml",
         rename_config=OUT_DIR / "project_level_qc" / "multiqc" / "aggregated_rename_configs.tsv",
     output:
-        OUT_DIR / "project_level_qc" / "multiqc" / "multiqc_report.html",
+        protected(OUT_DIR / "project_level_qc" / "multiqc" / "multiqc_report.html"),
     message:
         "Running multiqc for all samples"
     params:

workflow/rules/coverage_analysis.smk

@@ -3,7 +3,7 @@ rule samtools_stats:
     input:
         PROJECT_PATH / "{sample}" / "bam" / "{sample}.bam",
     output:
-        OUT_DIR / "{sample}" / "qc" / "samtools-stats" / "{sample}.txt",
+        protected(OUT_DIR / "{sample}" / "qc" / "samtools-stats" / "{sample}.txt"),
     wrapper:
         "0.64.0/bio/samtools/stats"
 
@@ -15,9 +15,9 @@ rule qualimap_bamqc:
         index=PROJECT_PATH / "{sample}" / "bam" / "{sample}.bam.bai",
         target_regions=get_capture_regions_bed,
     output:
-        html_report=OUT_DIR / "{sample}" / "qc" / "qualimap" / "{sample}" / "qualimapReport.html",
-        coverage=OUT_DIR / "{sample}" / "qc/qualimap/{sample}/raw_data_qualimapReport/coverage_across_reference.txt",
-        summary=OUT_DIR / "{sample}" / "qc" / "qualimap" / "{sample}" / "genome_results.txt",
+        html_report=protected(OUT_DIR / "{sample}" / "qc" / "qualimap" / "{sample}" / "qualimapReport.html"),
+        coverage=protected(OUT_DIR / "{sample}" / "qc/qualimap/{sample}/raw_data_qualimapReport/coverage_across_reference.txt"),
+        summary=protected(OUT_DIR / "{sample}" / "qc" / "qualimap" / "{sample}" / "genome_results.txt"),
     message:
         "stats bam using qualimap"
     conda:
@@ -47,8 +47,8 @@ rule mosdepth_coverage:
         bam_index=PROJECT_PATH / "{sample}" / "bam" / "{sample}.bam.bai",
         target_regions=get_capture_regions_bed,
     output:
-        dist=OUT_DIR / "{sample}" / "qc" / "mosdepth" / "{sample}.mosdepth.global.dist.txt",
-        summary=OUT_DIR / "{sample}" / "qc" / "mosdepth" / "{sample}.mosdepth.summary.txt",
+        dist=protected(OUT_DIR / "{sample}" / "qc" / "mosdepth" / "{sample}.mosdepth.global.dist.txt"),
+        summary=protected(OUT_DIR / "{sample}" / "qc" / "mosdepth" / "{sample}.mosdepth.summary.txt"),
     message:
         "Running mosdepth for coverage. Sample: {wildcards.sample}"
     conda:
@@ -74,7 +74,7 @@ rule mosdepth_plot:
         dist=expand(OUT_DIR / "{sample}" / "qc" / "mosdepth" / "{sample}.mosdepth.global.dist.txt", sample=SAMPLES),
         script=WORKFLOW_PATH / "src/mosdepth/v0.3.1/plot-dist.py",
     output:
-        OUT_DIR / "project_level_qc" / "mosdepth" / "mosdepth.html",
+        protected(OUT_DIR / "project_level_qc" / "mosdepth" / "mosdepth.html"),
     message:
         "Running mosdepth plotting"
     params:
@@ -99,8 +99,8 @@ rule indexcov:
             sample=SAMPLES,
         ),
     output:
-        html=OUT_DIR / "project_level_qc" / "indexcov" / "index.html",
-        bed=OUT_DIR / "project_level_qc" / "indexcov" / "indexcov-indexcov.bed.gz",
+        html=protected(OUT_DIR / "project_level_qc" / "indexcov" / "index.html"),
+        bed=protected(OUT_DIR / "project_level_qc" / "indexcov" / "indexcov-indexcov.bed.gz"),
     message:
         "Running indexcov"
     log:
@@ -125,7 +125,7 @@ rule covviz:
         bed=OUT_DIR / "project_level_qc" / "indexcov" / "indexcov-indexcov.bed.gz",
         ped=PEDIGREE_FPATH,
     output:
-        html=OUT_DIR / "project_level_qc" / "covviz" / "covviz_report.html",
+        html=protected(OUT_DIR / "project_level_qc" / "covviz" / "covviz_report.html"),
     message:
         "Running covviz"
     log:

workflow/rules/relatedness_ancestry.smk

@@ -5,7 +5,7 @@ rule somalier_extract:
         sites=config["somalier"]["sites"],
         ref_genome=config["ref"],
     output:
-        OUT_DIR / "project_level_qc" / "somalier" / "extract" / "{sample}.somalier",
+        protected(OUT_DIR / "project_level_qc" / "somalier" / "extract" / "{sample}.somalier"),
     message:
         "Running somalier extract. Sample: {wildcards.sample}"
     singularity:
@@ -27,10 +27,10 @@ rule somalier_relate:
         extracted=expand(OUT_DIR / "project_level_qc" / "somalier" / "extract" / "{sample}.somalier", sample=SAMPLES),
         ped=PEDIGREE_FPATH,
     output:
-        out=expand(
+        out=protected(expand(
             OUT_DIR / "project_level_qc" / "somalier" / "relatedness" / "somalier.{ext}",
             ext=["html", "pairs.tsv", "samples.tsv"],
-        ),
+        )),
     message:
         "Running somalier relate"
     singularity:
@@ -59,10 +59,10 @@ rule somalier_ancestry:
         labels_1kg=config["somalier"]["labels_1kg"],
         somalier_1kg=directory(config["somalier"]["somalier_1kg"]),
     output:
-        out=expand(
+        out=protected(expand(
             OUT_DIR / "project_level_qc" / "somalier" / "ancestry" / "somalier.somalier-ancestry.{ext}",
             ext=["html", "tsv"],
-        ),
+        )),
     message:
         "Running somalier ancestry."
     singularity:

workflow/rules/vcf_stats.smk

@@ -2,7 +2,7 @@ rule bcftools_stats:
     input:
         PROJECT_PATH / "{sample}" / "vcf" / "{sample}.vcf.gz",
     output:
-        OUT_DIR / "{sample}" / "qc" / "bcftools-stats" / "{sample}.bcftools.stats",
+        protected(OUT_DIR / "{sample}" / "qc" / "bcftools-stats" / "{sample}.bcftools.stats"),
     message:
         "stats vcf using bcftools"
     conda:

workflow/rules/within_species_contamintation.smk

@@ -12,8 +12,8 @@ rule verifybamid:
         ref_genome=config["ref"],
         svd=get_svd,
     output:
-        ancestry=OUT_DIR / "{sample}" / "qc" / "verifyBamID" / "{sample}.Ancestry",
-        selfsm=OUT_DIR / "{sample}" / "qc" / "verifyBamID" / "{sample}.selfSM",
+        ancestry=protected(OUT_DIR / "{sample}" / "qc" / "verifyBamID" / "{sample}.Ancestry"),
+        selfsm=protected(OUT_DIR / "{sample}" / "qc" / "verifyBamID" / "{sample}.selfSM"),
     message:
         "Running VerifyBamID to detect within-species contamination. sample: {wildcards.sample}"
     conda: