draft of tstv

ktmeaton · Aug 24, 2021 · 9b80539 · 9b80539
1 parent 5280263
commit 9b80539
Show file tree

Hide file tree

Showing 3 changed files with 49 additions and 1 deletion.
diff --git a/workflow/rules/phylogeny.smk b/workflow/rules/phylogeny.smk
@@ -90,6 +90,7 @@ rule iqtree_stats:
         tree        = results_dir + "/iqtree/{reads_origin}/{locus_name}/{prune}/filter{missing_data}/filter-taxa/iqtree.treefile",
         tsv         = results_dir + "/iqtree/{reads_origin}/{locus_name}/{prune}/filter{missing_data}/filter-taxa/metadata.tsv",
         dnds        = results_dir + "/dnds_collect/{reads_origin}/{locus_name}/dnds.txt",
+        tstv        = results_dir + "/tstv_collect/{reads_origin}/{locus_name}/tstv.txt",
     output:
         constraints = results_dir + "/iqtree/{reads_origin}/{locus_name}/{prune}/filter{missing_data}/filter-taxa/temporal_constraints.txt",
     log:

diff --git a/workflow/rules/qc.smk b/workflow/rules/qc.smk
@@ -160,7 +160,7 @@ rule tstv:
         vcf = results_dir + "/snippy_pairwise/{reads_origin}/{sample}/{sample}.subs.vcf",
     output:
         vcf  = results_dir + "/tstv/{reads_origin}/{locus_name}/{sample}.vcf",
-        tstv = results_dir + "/tstv/{reads_origin}/{locus_name}/{sample}.tstv",
+        tstv = results_dir + "/tstv/{reads_origin}/{locus_name}/{sample}.txt",
     params:
         locus = config["reference_locus"],
     shell:
@@ -174,3 +174,23 @@ rule tstv:
         {scripts_dir}/vcf_tstv.sh {output.tstv}.raw > {output.tstv}
         rm -f {output.tstv}.raw
         """
+
+rule tstv_collect:
+    """
+    Collect tStV statistics.
+    """
+    input:
+        files = lambda wildcards: remove_duplicates([
+                os.path.join(
+                    os.path.dirname(os.path.dirname(path)),
+                    config["reference_locus_name"],
+                    os.path.basename(os.path.dirname(path)) + ".txt")
+                for path in identify_paths(outdir="tstv", reads_origin=wildcards.reads_origin)]),
+    output:
+        df = results_dir + "/tstv_collect/{reads_origin}/{locus_name}/tstv.txt",
+
+    shell:
+        """
+        echo -e "sample\ttstv" > {output.df};
+        for file in {input.files}; do head -n1 $file; done >> {output.df};
+        """
diff --git a/workflow/rules/targets.smk b/workflow/rules/targets.smk
@@ -117,6 +117,33 @@ rule dnds_collect_local:
     input:
         dnds_collect_local_input
 
+#------------------------------------------------------------------------------#
+# TsTv
+#------------------------------------------------------------------------------#
+
+tstv_collect_all_input = results_dir + expand("/tstv_collect/all/{locus_name}/tstv.txt",
+                                locus_name=config["reference_locus_name"])[0]
+tstv_collect_local_input    = tstv_collect_all_input.replace("all", "local")
+tstv_collect_assembly_input    = tstv_collect_all_input.replace("all", "assembly")
+tstv_collect_sra_input    = tstv_collect_all_input.replace("all", "sra")
+
+rule tstv_collect_all:
+    input:
+        tstv_collect_all_input
+
+rule tstv_collect_assembly:
+    input:
+        tstv_collect_assembly_input
+
+rule tstv_collect_sra:
+    input:
+        tstv_collect_sra_input,
+
+rule tstv_collect_local:
+    input:
+        tstv_collect_local_input
+
+
 #------------------------------------------------------------------------------#
 # Filtering
 #------------------------------------------------------------------------------#