Cleaned up parsing and added ref stats

mc_msa.py

@@ -48,17 +48,18 @@ def create_config() :
             actual_regions = "["
             last_elem = None
             for elem in regions:
+                qt = "" if "'" in elem else "'"
                 if '_' in elem:
-                    actual_regions+= elem +","
+                    actual_regions+= qt+ elem.replace('"', "'").replace(" ", "") +qt +","
                 elif args.size != 0:
-                    actual_regions+= elem+"_"+str(int(elem)+int(args.size))+","
+                    actual_regions+= qt+ elem.replace('"', "'").replace(" ", "")+"_"+str(int(elem)+int(args.size))+ qt+ ","
                 elif last_elem:
-                    actual_regions+= last_elem+'_'+elem+","
+                    actual_regions+= qt+last_elem+'_'+elem.replace('"', "'").replace(" ", "")+qt+","
                     last_elem = elem
                 else:
                     last_elem = elem
-            print(actual_regions)
-            c_file.write("region: "+actual_regions[:-1]+"]\n")
+            actual_regions = actual_regions[:-1]
+            c_file.write("region: "+actual_regions+"]\n")
         else:
             c_file.write("region_size: " + args.size+"\n")
             c_file.write("region_overlap: "+ args.region_overlap+"\n")

workflow/Snakefile

@@ -554,6 +554,12 @@ if "ref_file" in config:
     conda: "envs/exonerate.yaml"
     shell: "exonerate --bestn 1 -Q dna -E -m a:g --showvulgar false --showsugar false --showcigar false --showalignment true {input.target} {input.query} > {output} 2>/dev/null"
 
+   rule stats_ref_metaconsensus:
+    input: query=rules.concat_region.output, target=config["ref_file"]
+    output: os.path.join("{output_folder}", "plot", "stats_meta_ref_d{depth}_t1_{threshold}_t2_{value}.fasta")
+    conda: "envs/exonerate.yaml"
+    shell: "./workflow/scripts/exonerate_stats.sh {output} {input.target} {input.query}"
+
    rule extract_mismatches:
     input: rules.comparison_ref_metaconsensus.output
     output: os.path.join("{output_folder}", "mismatches", "mismatches_d{depth}_t1_{threshold}_t2_{value}.fasta")
@@ -565,7 +571,8 @@ if "ref_file" in config:
             expand(rules.plot_t1.output, region=config["region"], depth=config["depth"] , plot_folder=config["output_folder"]),
             expand(rules.plot_depth.output, region=config["region"], threshold=config["threshold_1"], plot_folder=config["output_folder"]),
             expand(rules.plot_depth_all.output, region=config["region"], threshold=config["threshold_1"], plot_folder=config["output_folder"]),
-            expand(rules.extract_mismatches.output , threshold=config["threshold_1"], value=config["threshold_2"], output_folder=config["output_folder"], depth=config["depth"])
+            expand(rules.extract_mismatches.output , threshold=config["threshold_1"], value=config["threshold_2"], output_folder=config["output_folder"], depth=config["depth"]),
+            expand(rules.stats_ref_metaconsensus.output , threshold=config["threshold_1"], value=config["threshold_2"], output_folder=config["output_folder"], depth=config["depth"])
 
      message: "Created all plots"