refactor: filter pairwise correlations with a minimum number of unique observations

Author: ahmig

workflow/rules/vaf.smk

@@ -308,6 +308,7 @@ rule pairwise_trajectory_correlation_data:
     params:
         cor_method=config["COR"]["METHOD"],
         cor_use="pairwise.complete.obs",
+        n_threshold=0,
     input:
         variants="<results>/<dataset>/variants.all_sites.tsv",
         metadata=config["METADATA"],

workflow/scripts/report/pairwise_trajectory_correlation_data.R

@@ -45,9 +45,13 @@ all_variants_wider <- variants %>%
 log_info("Saving table of frequencies")
 write.csv(all_variants_wider, snakemake@output[["table"]])
 
+log_info("Filtering out variant columns regarding unique values")
+filtered <- all_variants_wider %>%
+  select(where(~ length(unique(na.omit(.))) > snakemake@params$n_threshold))
+
 log_info("Calculating correlations")
 cor.mat <- cor(
-  all_variants_wider,
+  filtered,
   method = snakemake@params$cor_method,
   use = snakemake@params$cor_use
 )