refactor: filter_sets & filter_list selection moved to Snakefile

bihealth · Apr 24, 2024 · ea59102 · ea59102
1 parent e2dba3d
commit ea59102
Showing 1 changed file with 177 additions and 184 deletions.
diff --git a/snappy_pipeline/workflows/somatic_variant_filtration/Snakefile b/snappy_pipeline/workflows/somatic_variant_filtration/Snakefile
@@ -53,27 +53,6 @@ rule somatic_variant_filtration_link_out_run:
 
 # Somatic Variant Filtration ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
 
-# Run DKFZ Bias Filter --------------------------------------------------------
-
-
-rule somatic_variant_filtration_dkfz_bias_filter_run:
-    input:
-        **wf.get_input_files("dkfz_bias_filter", "run"),
-    output:
-        **wf.get_output_files("dkfz_bias_filter", "run"),
-    threads: wf.get_resource("dkfz_bias_filter", "run", "threads")
-    resources:
-        time=wf.get_resource("dkfz_bias_filter", "run", "time"),
-        memory=wf.get_resource("dkfz_bias_filter", "run", "memory"),
-        partition=wf.get_resource("dkfz_bias_filter", "run", "partition"),
-        tmpdir=wf.get_resource("dkfz_bias_filter", "run", "tmpdir"),
-    params:
-        **{"args": wf.get_params("dkfz_bias_filter", "run")},
-    log:
-        **wf.get_log_file("dkfz_bias_filter", "run"),
-    wrapper:
-        wf.wrapper_path("dkfz_bias_filter")
-
 
 # Write out "panel of normals" file for eb_filter ------------------------------
 
@@ -95,175 +74,189 @@ rule somatic_variant_filtration_eb_filter_write_panel:
         wf.substep_getattr("eb_filter", "write_panel_of_normals_file")(wildcards)
 
 
-# Run eb_filter ----------------------------------------------------------------
-
-
-rule somatic_variant_filtration_eb_filter_run:
-    input:
-        unpack(wf.get_input_files("eb_filter", "run")),
-    output:
-        **wf.get_output_files("eb_filter", "run"),
-    threads: wf.get_resource("eb_filter", "run", "threads")
-    resources:
-        time=wf.get_resource("eb_filter", "run", "time"),
-        memory=wf.get_resource("eb_filter", "run", "memory"),
-        partition=wf.get_resource("eb_filter", "run", "partition"),
-        tmpdir=wf.get_resource("eb_filter", "run", "tmpdir"),
-    params:
-        **{"args": wf.get_params("eb_filter", "run")},
-    log:
-        **wf.get_log_file("eb_filter", "run"),
-    wrapper:
-        wf.wrapper_path("eb_filter")
-
-
-# Apply Filters ---------------------------------------------------------------
-
-
-rule variant_filtration_apply_filters_run:
-    input:
-        **(wf.get_input_files("apply_filters", "run")),
-    output:
-        **wf.get_output_files("apply_filters", "run"),
-    threads: wf.get_resource("apply_filters", "run", "threads")
-    resources:
-        time=wf.get_resource("apply_filters", "run", "time"),
-        memory=wf.get_resource("apply_filters", "run", "memory"),
-        partition=wf.get_resource("apply_filters", "run", "partition"),
-        tmpdir=wf.get_resource("apply_filters", "run", "tmpdir"),
-    log:
-        wf.get_log_file("apply_filters", "run"),
-    params:
-        args=wf.substep_dispatch("apply_filters", "get_args", "run"),
-    wrapper:
-        wf.wrapper_path("somatic_variant_filtration/apply_filters")
-
-
-# Filter to Exons--------------------------------------------------------------
-
+# Run DKFZ Bias Filter --------------------------------------------------------
 
-rule variant_filtration_filter_to_exons_run:
-    input:
-        unpack(wf.get_input_files("filter_to_exons", "run")),
-    output:
-        **wf.get_output_files("filter_to_exons", "run"),
-    threads: wf.get_resource("filter_to_exons", "run", "threads")
-    resources:
-        time=wf.get_resource("filter_to_exons", "run", "time"),
-        memory=wf.get_resource("filter_to_exons", "run", "memory"),
-        partition=wf.get_resource("filter_to_exons", "run", "partition"),
-        tmpdir=wf.get_resource("filter_to_exons", "run", "tmpdir"),
-    log:
-        wf.get_log_file("filter_to_exons", "run"),
-    wrapper:
-        wf.wrapper_path("somatic_variant_filtration/filter_to_exons")
+if len(config["step_config"]["somatic_variant_filtration"]["filter_sets"]) > 0:
+
+    rule somatic_variant_filtration_dkfz_bias_filter_run:
+        input:
+            **wf.get_input_files("dkfz_bias_filter", "run"),
+        output:
+            **wf.get_output_files("dkfz_bias_filter", "run"),
+        threads: wf.get_resource("dkfz_bias_filter", "run", "threads")
+        resources:
+            time=wf.get_resource("dkfz_bias_filter", "run", "time"),
+            memory=wf.get_resource("dkfz_bias_filter", "run", "memory"),
+            partition=wf.get_resource("dkfz_bias_filter", "run", "partition"),
+            tmpdir=wf.get_resource("dkfz_bias_filter", "run", "tmpdir"),
+        params:
+            **{"args": wf.get_params("dkfz_bias_filter", "run")},
+        log:
+            **wf.get_log_file("dkfz_bias_filter", "run"),
+        wrapper:
+            wf.wrapper_path("dkfz_bias_filter")
+
+    # Run eb_filter ----------------------------------------------------------------
+
+    rule somatic_variant_filtration_eb_filter_run:
+        input:
+            unpack(wf.get_input_files("eb_filter", "run")),
+        output:
+            **wf.get_output_files("eb_filter", "run"),
+        threads: wf.get_resource("eb_filter", "run", "threads")
+        resources:
+            time=wf.get_resource("eb_filter", "run", "time"),
+            memory=wf.get_resource("eb_filter", "run", "memory"),
+            partition=wf.get_resource("eb_filter", "run", "partition"),
+            tmpdir=wf.get_resource("eb_filter", "run", "tmpdir"),
+        params:
+            **{"args": wf.get_params("eb_filter", "run")},
+        log:
+            **wf.get_log_file("eb_filter", "run"),
+        wrapper:
+            wf.wrapper_path("eb_filter")
+
+    # Apply Filters ---------------------------------------------------------------
+
+    rule variant_filtration_apply_filters_run:
+        input:
+            **(wf.get_input_files("apply_filters", "run")),
+        output:
+            **wf.get_output_files("apply_filters", "run"),
+        threads: wf.get_resource("apply_filters", "run", "threads")
+        resources:
+            time=wf.get_resource("apply_filters", "run", "time"),
+            memory=wf.get_resource("apply_filters", "run", "memory"),
+            partition=wf.get_resource("apply_filters", "run", "partition"),
+            tmpdir=wf.get_resource("apply_filters", "run", "tmpdir"),
+        log:
+            wf.get_log_file("apply_filters", "run"),
+        params:
+            args=wf.substep_dispatch("apply_filters", "get_args", "run"),
+        wrapper:
+            wf.wrapper_path("somatic_variant_filtration/apply_filters")
+
+    # Filter to Exons--------------------------------------------------------------
+
+    rule variant_filtration_filter_to_exons_run:
+        input:
+            unpack(wf.get_input_files("filter_to_exons", "run")),
+        output:
+            **wf.get_output_files("filter_to_exons", "run"),
+        threads: wf.get_resource("filter_to_exons", "run", "threads")
+        resources:
+            time=wf.get_resource("filter_to_exons", "run", "time"),
+            memory=wf.get_resource("filter_to_exons", "run", "memory"),
+            partition=wf.get_resource("filter_to_exons", "run", "partition"),
+            tmpdir=wf.get_resource("filter_to_exons", "run", "tmpdir"),
+        log:
+            wf.get_log_file("filter_to_exons", "run"),
+        wrapper:
+            wf.wrapper_path("somatic_variant_filtration/filter_to_exons")
 
 
 # Flexible Somatic Variant Filtration ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
 
 
-checkpoint one_dkfz:
-    input:
-        unpack(wf.get_input_files("one_dkfz", "run")),
-    output:
-        **wf.get_output_files("one_dkfz", "run"),
-    threads: wf.get_resource("one_dkfz", "run", "threads")
-    resources:
-        time=wf.get_resource("one_dkfz", "run", "time"),
-        memory=wf.get_resource("one_dkfz", "run", "memory"),
-        partition=wf.get_resource("one_dkfz", "run", "partition"),
-        tmpdir=wf.get_resource("one_dkfz", "run", "tmpdir"),
-    log:
-        **wf.get_log_file("one_dkfz", "run"),
-    wrapper:
-        wf.wrapper_path("dkfz_bias_filter")
-
-
-checkpoint one_ebfilter:
-    input:
-        unpack(wf.get_input_files("one_ebfilter", "run")),
-    output:
-        **wf.get_output_files("one_ebfilter", "run"),
-    params:
-        wf.get_params("one_ebfilter", "run"),
-    threads: wf.get_resource("one_ebfilter", "run", "threads")
-    resources:
-        time=wf.get_resource("one_ebfilter", "run", "time"),
-        memory=wf.get_resource("one_ebfilter", "run", "memory"),
-        partition=wf.get_resource("one_ebfilter", "run", "partition"),
-        tmpdir=wf.get_resource("one_ebfilter", "run", "tmpdir"),
-    log:
-        **wf.get_log_file("one_ebfilter", "run"),
-    wrapper:
-        wf.wrapper_path("eb_filter")
-
-
-checkpoint one_bcftools:
-    input:
-        unpack(wf.get_input_files("one_bcftools", "run")),
-    output:
-        **wf.get_output_files("one_bcftools", "run"),
-    threads: wf.get_resource("one_bcftools", "run", "threads")
-    resources:
-        time=wf.get_resource("one_bcftools", "run", "time"),
-        memory=wf.get_resource("one_bcftools", "run", "memory"),
-        partition=wf.get_resource("one_bcftools", "run", "partition"),
-        tmpdir=wf.get_resource("one_bcftools", "run", "tmpdir"),
-    log:
-        **wf.get_log_file("one_bcftools", "run"),
-    wrapper:
-        wf.wrapper_path("bcftools/filter")
-
-
-checkpoint one_regions:
-    input:
-        unpack(wf.get_input_files("one_regions", "run")),
-    output:
-        **wf.get_output_files("one_regions", "run"),
-    threads: wf.get_resource("one_regions", "run", "threads")
-    resources:
-        time=wf.get_resource("one_regions", "run", "time"),
-        memory=wf.get_resource("one_regions", "run", "memory"),
-        partition=wf.get_resource("one_regions", "run", "partition"),
-        tmpdir=wf.get_resource("one_regions", "run", "tmpdir"),
-    log:
-        **wf.get_log_file("one_regions", "run"),
-    wrapper:
-        wf.wrapper_path("bcftools/regions")
-
-
-checkpoint one_protected:
-    input:
-        unpack(wf.get_input_files("one_protected", "run")),
-    output:
-        **wf.get_output_files("one_protected", "run"),
-    threads: wf.get_resource("one_protected", "run", "threads")
-    resources:
-        time=wf.get_resource("one_protected", "run", "time"),
-        memory=wf.get_resource("one_protected", "run", "memory"),
-        partition=wf.get_resource("one_protected", "run", "partition"),
-        tmpdir=wf.get_resource("one_protected", "run", "tmpdir"),
-    log:
-        **wf.get_log_file("one_protected", "run"),
-    wrapper:
-        wf.wrapper_path("bcftools/protected")
-
-
-rule last_filter:
-    input:
-        **wf.get_input_files("last_filter", "run"),
-    output:
-        **wf.get_output_files("last_filter", "run"),
-    threads: wf.get_resource("last_filter", "run", "threads")
-    resources:
-        time=wf.get_resource("last_filter", "run", "time"),
-        memory=wf.get_resource("last_filter", "run", "memory"),
-        partition=wf.get_resource("last_filter", "run", "partition"),
-        tmpdir=wf.get_resource("last_filter", "run", "tmpdir"),
-    log:
-        **wf.get_log_file("last_filter", "run"),
-    wrapper:
-        wf.wrapper_path("somatic_variant_filtration/apply_all_filters")
+if len(config["step_config"]["somatic_variant_filtration"]["filter_list"]) > 0:
+
+    checkpoint one_dkfz:
+        input:
+            unpack(wf.get_input_files("one_dkfz", "run")),
+        output:
+            **wf.get_output_files("one_dkfz", "run"),
+        threads: wf.get_resource("one_dkfz", "run", "threads")
+        resources:
+            time=wf.get_resource("one_dkfz", "run", "time"),
+            memory=wf.get_resource("one_dkfz", "run", "memory"),
+            partition=wf.get_resource("one_dkfz", "run", "partition"),
+            tmpdir=wf.get_resource("one_dkfz", "run", "tmpdir"),
+        log:
+            **wf.get_log_file("one_dkfz", "run"),
+        wrapper:
+            wf.wrapper_path("dkfz_bias_filter")
+
+    checkpoint one_ebfilter:
+        input:
+            unpack(wf.get_input_files("one_ebfilter", "run")),
+        output:
+            **wf.get_output_files("one_ebfilter", "run"),
+        params:
+            wf.get_params("one_ebfilter", "run"),
+        threads: wf.get_resource("one_ebfilter", "run", "threads")
+        resources:
+            time=wf.get_resource("one_ebfilter", "run", "time"),
+            memory=wf.get_resource("one_ebfilter", "run", "memory"),
+            partition=wf.get_resource("one_ebfilter", "run", "partition"),
+            tmpdir=wf.get_resource("one_ebfilter", "run", "tmpdir"),
+        log:
+            **wf.get_log_file("one_ebfilter", "run"),
+        wrapper:
+            wf.wrapper_path("eb_filter")
+
+    checkpoint one_bcftools:
+        input:
+            unpack(wf.get_input_files("one_bcftools", "run")),
+        output:
+            **wf.get_output_files("one_bcftools", "run"),
+        threads: wf.get_resource("one_bcftools", "run", "threads")
+        resources:
+            time=wf.get_resource("one_bcftools", "run", "time"),
+            memory=wf.get_resource("one_bcftools", "run", "memory"),
+            partition=wf.get_resource("one_bcftools", "run", "partition"),
+            tmpdir=wf.get_resource("one_bcftools", "run", "tmpdir"),
+        log:
+            **wf.get_log_file("one_bcftools", "run"),
+        wrapper:
+            wf.wrapper_path("bcftools/filter")
+
+    checkpoint one_regions:
+        input:
+            unpack(wf.get_input_files("one_regions", "run")),
+        output:
+            **wf.get_output_files("one_regions", "run"),
+        threads: wf.get_resource("one_regions", "run", "threads")
+        resources:
+            time=wf.get_resource("one_regions", "run", "time"),
+            memory=wf.get_resource("one_regions", "run", "memory"),
+            partition=wf.get_resource("one_regions", "run", "partition"),
+            tmpdir=wf.get_resource("one_regions", "run", "tmpdir"),
+        log:
+            **wf.get_log_file("one_regions", "run"),
+        wrapper:
+            wf.wrapper_path("bcftools/regions")
+
+    checkpoint one_protected:
+        input:
+            unpack(wf.get_input_files("one_protected", "run")),
+        output:
+            **wf.get_output_files("one_protected", "run"),
+        threads: wf.get_resource("one_protected", "run", "threads")
+        resources:
+            time=wf.get_resource("one_protected", "run", "time"),
+            memory=wf.get_resource("one_protected", "run", "memory"),
+            partition=wf.get_resource("one_protected", "run", "partition"),
+            tmpdir=wf.get_resource("one_protected", "run", "tmpdir"),
+        log:
+            **wf.get_log_file("one_protected", "run"),
+        wrapper:
+            wf.wrapper_path("bcftools/protected")
+
+    rule last_filter:
+        input:
+            **wf.get_input_files("last_filter", "run"),
+        output:
+            **wf.get_output_files("last_filter", "run"),
+        threads: wf.get_resource("last_filter", "run", "threads")
+        resources:
+            time=wf.get_resource("last_filter", "run", "time"),
+            memory=wf.get_resource("last_filter", "run", "memory"),
+            partition=wf.get_resource("last_filter", "run", "partition"),
+            tmpdir=wf.get_resource("last_filter", "run", "tmpdir"),
+        log:
+            **wf.get_log_file("last_filter", "run"),
+        wrapper:
+            wf.wrapper_path("somatic_variant_filtration/apply_all_filters")
 
 
 # Variant Statistics Computation ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~