fix: deduplicate variants and handle missing annotation

workflow/rules/common.smk

@@ -29,7 +29,7 @@ def get_repo_version(base_dir: str, default: str, warn=False) -> str:
     """
     try:
         last_tag_description = subprocess.check_output(
-            f"git --git-dir={base_dir}/.git describe --always",
+            f"git --git-dir={base_dir}/.git describe --tags --always",
             shell=True,
             stderr=subprocess.DEVNULL
         ).strip().decode("utf-8")

workflow/rules/context.smk

@@ -28,6 +28,7 @@ rule download_context:
         duplicate_accids = OUTDIR/"context"/"duplicate_accession_ids.txt"
     log:
         LOGDIR / "download_context" / "log.txt"
+    retries: 2
     script:
         "../scripts/download_context.R"
 

workflow/scripts/format_vcf_fields_longer.R

@@ -45,6 +45,9 @@ read_tsv(
   # Rename "...[*]..." columns using the provided lookup via Snakemake config
   rename(all_of(unlist(snakemake@params$colnames_mapping))) %>%
 
+  # Ensure missing values are properly encoded
+  mutate(across(where(is.character), ~ na_if(.x, "NA"))) %>%
+
   # Separate &-delimited error column (more than one error/warning/info message per row is possible)
   mutate(split_errors = strsplit(ERRORS, "&")) %>%
   # Keep rows with none of the excluded ERRORS terms, if any

workflow/scripts/report/pairwise_trajectory_correlation_data.R

@@ -25,7 +25,7 @@ date_order <- read_csv(snakemake@input[["metadata"]]) %>%
 
 log_info("Formatting variants")
 all_variants_wider <- variants %>%
-  select(SAMPLE, VARIANT_NAME, ALT_FREQ) %>%
+  distinct(SAMPLE, VARIANT_NAME, ALT_FREQ) %>%
   pivot_wider(
     names_from = VARIANT_NAME,
     values_from = ALT_FREQ