cleanup

modules/CREATE_SITECOUNT_MATRICES/Snakefile

@@ -97,19 +97,16 @@ def CSM_get_abspath(p):
     """
     return(p if p[0] == os.sep else os.path.abspath(p))
 
-
-
-
-
-def CSM_helper(wildcards):
+def CSM_region_helper(wildcards):
+    """
+    Helper function to determine wheter a region is supposed to be
+    analyzed in full or over windows
+    """
     if wildcards.CSM_region_id in config["CSM_regions_files"]:
         return config["CSM_regions_files"][wildcards.CSM_region_id]
     else:
         return config["CSM_additional_regions_files"][wildcards.CSM_region_id]
 
-
-
-
 ##############################################################################
 ### Target rule with final output of the pipeline
 ##############################################################################
@@ -201,7 +198,7 @@ rule CSM_extract_window_coord_and_sequence:
             "{CSM_outdir}",
             "CSM_outdir"
         ),
-        BED_region = lambda wildcards: CSM_helper(wildcards),
+        BED_region = lambda wildcards: CSM_region_helper(wildcards),
         FASTA_genome = config["CSM_genomic_sequence"],
         SCRIPT_ = os.path.join(
             config["CSM_scripts_dir"],

modules/CREATE_SITECOUNT_MATRICES/configs/config.yml

@@ -37,9 +37,7 @@ CSM_regions_files:
 
 # paths to the bed-formatted additional whole regions
 CSM_additional_regions_files:
-  "3pUTR_minlen150_prefix": "/data/scc2/ag-gruber-ext/wsciekly.maciek/mapp-stability-analysis/singular-sites-margin25-minutrlen150/clean-sites-prefix-UTR.bed"
-  "3pUTR_minlen150_total": "/data/scc2/ag-gruber-ext/wsciekly.maciek/mapp-stability-analysis/singular-sites-margin25-minutrlen150/clean-sites-total-UTR.bed"
-  "3pUTR_minlen150_suffix": "/data/scc2/ag-gruber-ext/wsciekly.maciek/mapp-stability-analysis/singular-sites-margin25-minutrlen150/clean-sites-suffix-UTR.bed"
+  "3pUTR_minlen150_total": "/data/scc2/ag-gruber-ext/MAPP_paper/supplementary_package/mapp-stability-analysis-BACKUP/singular-sites-margin25-minutrlen150/clean-sites-total-UTR.bed"
 
 # sliding window configuration: window size and slide step
 CSM_window_size: "50"
@@ -51,7 +49,7 @@ CSM_matrix_type: "kmers"
 # options for "kmers" sitecount matrices:
 # inclusive ends specification for the range of k-mers sizes = [kmer_min,kmer_max]
 CSM_kmer_min: "3"
-CSM_kmer_max: "3"
+CSM_kmer_max: "5"
 
 # options for "pwms" sitecount matrices:
 # path to the directory with TRANSFAC-formatted PWM files

modules/CREATE_SITECOUNT_MATRICES/profiles/local-conda/config.yaml

@@ -1,5 +1,5 @@
 snakefile: "../Snakefile"
-cores: 64
+cores: 1
 printshellcmds: true
 rerun-incomplete: true
 use-conda: true