fix: correct gtdb test using downloaded table

.tests/unit/gtdb_prep/data/config/lactobacillus_delbruecki/samples.csv

@@ -1,5 +1,5 @@
 genome_id,source,organism,genus,species,strain,closest_placement_reference,input_file
-GCA_000056065.1,ncbi,,,,,,
-GCA_000182835.1,ncbi,,,,,,
+GCF_000056065.1,ncbi,,,,,,
+GCF_000182835.1,ncbi,,,,,,
 GCA_000191165.1,ncbi,,,,,,
-GCA_000014405.1,ncbi,,,,,,
+GCF_000014405.1,ncbi,,,,,,

.tests/unit/gtdb_prep/expected/data/interim/gtdb/GCF_000056065.1.json

@@ -0,0 +1,140 @@
+{
+  "genome_id": "GCF_000056065.1",
+  "gtdb_url": "https://api.gtdb.ecogenomic.org/genome/GCF_000056065.1/taxon-history",
+  "gtdb_release": "R214",
+  "gtdb_taxonomy": {
+    "domain": "d__Bacteria",
+    "phylum": "p__Bacillota",
+    "class": "c__Bacilli",
+    "order": "o__Lactobacillales",
+    "family": "f__Lactobacillaceae",
+    "genus": "g__Lactobacillus",
+    "species": "s__Lactobacillus delbrueckii"
+  },
+  "metadata_url": "https://api.gtdb.ecogenomic.org/genome/GCF_000056065.1/card",
+  "metadata": {
+    "genome": {
+      "accession": "GCF_000056065.1",
+      "name": "GCF_000056065.1"
+    },
+    "metadata_nucleotide": {
+      "trna_aa_count": 19,
+      "contig_count": 1,
+      "n50_contigs": 1864998,
+      "longest_contig": 1864998,
+      "scaffold_count": 1,
+      "n50_scaffolds": 1864998,
+      "longest_scaffold": 1864998,
+      "genome_size": 1864998,
+      "gc_percentage": 49.71903455124348,
+      "ambiguous_bases": 0,
+      "gc_count": 927259,
+      "l50_contigs": 1,
+      "l50_scaffolds": 1,
+      "mean_contig_length": 1864998,
+      "mean_scaffold_length": 1864998
+    },
+    "metadata_gene": {
+      "checkm_completeness": 98.38,
+      "checkm_contamination": 0.0,
+      "checkm_strain_heterogeneity": 0.0,
+      "lsu_5s_count": 9,
+      "ssu_count": 9,
+      "lsu_23s_count": 9,
+      "protein_count": 1916,
+      "coding_density": 84.01837428243891,
+      "lsu_23s_contig_len": "1864998",
+      "lsu_23s_length": "2904",
+      "lsu_23s_query_id": "NC_008054.1",
+      "lsu_5s_contig_len": "1864998",
+      "lsu_5s_length": "110",
+      "lsu_5s_query_id": "NC_008054.1",
+      "lsu_silva_23s_blast_align_len": "2904",
+      "lsu_silva_23s_blast_bitscore": "5347",
+      "lsu_silva_23s_blast_evalue": "0",
+      "lsu_silva_23s_blast_perc_identity": "99.897",
+      "lsu_silva_23s_blast_subject_id": "LUGK01000125.188.3095",
+      "lsu_silva_23s_taxonomy": "Bacteria;Firmicutes;Bacilli;Lactobacillales;Lactobacillaceae;Lactobacillus;Lactobacillus delbrueckii subsp. bulgaricus",
+      "checkm_marker_count": 430,
+      "checkm_marker_lineage": "g__Lactobacillus (UID377)",
+      "checkm_marker_set_count": 154,
+      "coding_bases": 1566941,
+      "mimag_high_quality": "t",
+      "mimag_low_quality": "f",
+      "mimag_medium_quality": "f",
+      "ssu_contig_len": "1864998",
+      "ssu_gg_blast_align_len": "none",
+      "ssu_gg_blast_bitscore": "none",
+      "ssu_gg_blast_evalue": "none",
+      "ssu_gg_blast_perc_identity": "none",
+      "ssu_gg_blast_subject_id": "none",
+      "ssu_gg_taxonomy": "none",
+      "ssu_length": "1562",
+      "ssu_query_id": "NC_008054.1-#3",
+      "ssu_silva_blast_align_len": "1559",
+      "ssu_silva_blast_bitscore": "2880",
+      "ssu_silva_blast_evalue": "0",
+      "ssu_silva_blast_perc_identity": "100",
+      "ssu_silva_blast_subject_id": "CR954253.1359934.1361495",
+      "ssu_silva_taxonomy": "Bacteria;Firmicutes;Bacilli;Lactobacillales;Lactobacillaceae;Lactobacillus;Lactobacillus delbrueckii subsp. bulgaricus ATCC 11842 = JCM 1002",
+      "total_gap_length": 0,
+      "trna_count": 95,
+      "trna_selenocysteine_count": 0
+    },
+    "metadata_ncbi": {
+      "ncbi_genbank_assembly_accession": "GCA_000056065.1",
+      "ncbi_strain_identifiers": "ATCC 11842",
+      "ncbi_assembly_level": "Complete Genome",
+      "ncbi_assembly_name": "ASM5606v1",
+      "ncbi_assembly_type": "na",
+      "ncbi_bioproject": "PRJNA224116",
+      "ncbi_biosample": "SAMEA3138258",
+      "ncbi_country": "none",
+      "ncbi_date": "2006-05-26",
+      "ncbi_genome_category": "none",
+      "ncbi_genome_representation": "full",
+      "ncbi_isolate": "none",
+      "ncbi_isolation_source": "none",
+      "ncbi_lat_lon": "none",
+      "ncbi_molecule_count": 1,
+      "ncbi_protein_count": "1836",
+      "ncbi_refseq_category": "na",
+      "ncbi_seq_rel_date": "2006/05/26",
+      "ncbi_spanned_gaps": 0,
+      "ncbi_species_taxid": 1584,
+      "ncbi_ssu_count": "9",
+      "ncbi_submitter": "Genoscope",
+      "ncbi_taxid": 390333,
+      "ncbi_total_gap_length": 0,
+      "ncbi_translation_table": "11",
+      "ncbi_trna_count": "95",
+      "ncbi_unspanned_gaps": 0,
+      "ncbi_contig_count": "none",
+      "ncbi_contig_n50": "none",
+      "ncbi_ncrna_count": "0",
+      "ncbi_organism_name": "Lactobacillus delbrueckii subsp. bulgaricus ATCC 11842 = JCM 1002",
+      "ncbi_rrna_count": "27",
+      "ncbi_scaffold_count": "1",
+      "ncbi_scaffold_l50": "1",
+      "ncbi_scaffold_n50": "1864998",
+      "ncbi_scaffold_n75": "1864998",
+      "ncbi_scaffold_n90": "1864998",
+      "ncbi_total_length": 1864998,
+      "ncbi_ungapped_length": 1864998,
+      "ncbi_wgs_master": "none"
+    },
+    "metadata_type_material": {
+      "gtdb_type_designation_ncbi_taxa": "type strain of subspecies",
+      "gtdb_type_designation_ncbi_taxa_sources": "LPSN",
+      "gtdb_type_species_of_genus": "f"
+    },
+    "metadataTaxonomy": {
+      "ncbi_taxonomy": "d__Bacteria;p__Firmicutes;c__Bacilli;o__Lactobacillales;f__Lactobacillaceae;g__Lactobacillus;s__Lactobacillus delbrueckii",
+      "ncbi_taxonomy_unfiltered": "d__Bacteria;x__Terrabacteria group;p__Firmicutes;c__Bacilli;o__Lactobacillales;f__Lactobacillaceae;g__Lactobacillus;s__Lactobacillus delbrueckii;sb__Lactobacillus delbrueckii subsp. bulgaricus;x__Lactobacillus delbrueckii subsp. bulgaricus ATCC 11842 = JCM 1002",
+      "gtdb_representative": "f",
+      "gtdb_genome_representative": "RS_GCF_001433875.1",
+      "ncbi_type_material_designation": "assembly from type material"
+    },
+    "detail": "genome_found"
+  }
+}

workflow/rules/gtdb.smk

@@ -85,7 +85,7 @@ rule gtdb_prep:
             echo "gtdb_prep.py executed successfully"
         else
             echo "gtdb_prep.py failed, getting dataset from table instead..." >> {log}
-            if [ ! -f resources/gtdb_download/bac120_metadata_r{params.gtdb_release}.tsv ]; then
+            if [ ! -f resources/gtdb_download/bac120_metadata_{params.gtdb_release}.tsv ]; then
                 mkdir -p resources/gtdb_download/
                 wget -P resources/gtdb_download/ {params.gtdb_table} -nc 2>> {log}
                 gunzip -c resources/gtdb_download/bac120_metadata_{params.gtdb_release_version}.tsv.gz > resources/gtdb_download/bac120_metadata_{params.gtdb_release_version}.tsv 2>> {log}