Consistent naming of reference genome files

* Dropped "_dengue" from names, since it's redundant in the context of the project * Added "_genome" to the end of the reference genome files, to parallel the "_E" at the end of the reference E sequence files.
nextstrain · May 2, 2024 · ab5dff4 · ab5dff4
1 parent 2f48ddb
commit ab5dff4
Show file tree

Hide file tree

Showing 8 changed files with 6 additions and 6 deletions.
diff --git a/phylogenetic/config/reference_dengue_all.gb → phylogenetic/config/reference_all_genome.gb b/phylogenetic/config/reference_dengue_all.gb → phylogenetic/config/reference_all_genome.gb
diff --git a/...ogenetic/config/reference_dengue_denv1.gb → ...ogenetic/config/reference_denv1_genome.gb b/...ogenetic/config/reference_dengue_denv1.gb → ...ogenetic/config/reference_denv1_genome.gb
diff --git a/...ogenetic/config/reference_dengue_denv2.gb → ...ogenetic/config/reference_denv2_genome.gb b/...ogenetic/config/reference_dengue_denv2.gb → ...ogenetic/config/reference_denv2_genome.gb
diff --git a/...ogenetic/config/reference_dengue_denv3.gb → ...ogenetic/config/reference_denv3_genome.gb b/...ogenetic/config/reference_dengue_denv3.gb → ...ogenetic/config/reference_denv3_genome.gb
diff --git a/...ogenetic/config/reference_dengue_denv4.gb → ...ogenetic/config/reference_denv4_genome.gb b/...ogenetic/config/reference_dengue_denv4.gb → ...ogenetic/config/reference_denv4_genome.gb
diff --git a/phylogenetic/rules/annotate_phylogeny.smk b/phylogenetic/rules/annotate_phylogeny.smk
@@ -45,7 +45,7 @@ rule translate:
         tree = "results/tree_{serotype}_{gene}.nwk",
         node_data = "results/nt-muts_{serotype}_{gene}.json",
         # The genbank references for the E gene are dynamically generated files located within the results folder.
-        reference = lambda wildcard: "config/reference_dengue_{serotype}.gb" if wildcard.gene in ["genome"] else "results/config/reference_dengue_{serotype}_{gene}.gb"
+        reference = lambda wildcard: "config/reference_{serotype}_{gene}.gb" if wildcard.gene in ["genome"] else "results/config/reference_{serotype}_{gene}.gb"
     output:
         node_data = "results/aa-muts_{serotype}_{gene}.json"
     shell:

diff --git a/phylogenetic/rules/prepare_sequences.smk b/phylogenetic/rules/prepare_sequences.smk
@@ -82,7 +82,7 @@ rule align:
     """
     input:
         sequences = "results/filtered_{serotype}.fasta",
-        reference = "config/reference_dengue_{serotype}.gb"
+        reference = "config/reference_{serotype}_genome.gb"
     output:
         alignment = "results/aligned_{serotype}_genome.fasta"
     shell:

diff --git a/phylogenetic/rules/prepare_sequences_E.smk b/phylogenetic/rules/prepare_sequences_E.smk
@@ -22,10 +22,10 @@ rule generate_E_reference_files:
     Generating reference files for the E gene
     """
     input:
-        reference = "config/reference_dengue_{serotype}.gb",
+        reference = "config/reference_{serotype}_genome.gb",
     output:
-        fasta = "results/config/reference_dengue_{serotype}_E.fasta",
-        genbank = "results/config/reference_dengue_{serotype}_E.gb",
+        fasta = "results/config/reference_{serotype}_E.fasta",
+        genbank = "results/config/reference_{serotype}_E.gb",
     params:
         gene = "E",
     shell:
@@ -43,7 +43,7 @@ rule nextclade3_cut_E:
     """
     input:
         sequences = "data/sequences_{serotype}.fasta",
-        reference = "results/config/reference_dengue_{serotype}_E.fasta"
+        reference = "results/config/reference_{serotype}_E.fasta"
     output:
         sequences = "results/sequences_{serotype}_E.fasta"
     params: