added function to check values of consensus peak filtering settings

Author: niekwit

workflow/Snakefile

@@ -23,6 +23,7 @@ containerized: f"docker://niekwit/damid-seq:{VERSION}"
 # Load and validate config file
 configfile: "config/config.yaml"
 validate(config, "schemas/config.schema.yaml")
+check_consensus_peak_settings()
 
 plasmid_name = plasmid_fasta_name(config["plasmid_fasta"])
 

workflow/scripts/general_functions.smk

@@ -465,4 +465,15 @@ def input_extension():
     if paired_end:
         return []
     else:
-        return ".fastq.gz"
+        return ".fastq.gz"
+
+
+def check_consensus_peak_settings():
+    keep = config["consensus_peaks"]["keep"]
+
+    # Get number of subdirectories in reads/
+    subdirs = glob.glob("reads/*")
+    subdirs = len([d for d in subdirs if os.path.isdir(d)])
+
+    if keep > subdirs:
+        raise ValueError(f"Number of overlapping peaks to keep consensus peaks (config > consensus_peak > keep) is greater than number of subdirectories in reads/...")