add initial dnaapler archaea #75

src/dnaapler/__init__.py

@@ -204,6 +204,80 @@ def chromosome(
     end_dnaapler(start_time)
 
 
+"""
+archaea command
+"""
+
+@main_cli.command()
+@click.help_option("--help", "-h")
+@click.version_option(get_version(), "--version", "-V")
+@click.pass_context
+@common_options
+@click.option(
+    "-e",
+    "--evalue",
+    default="1e-10",
+    help="e value for blastx",
+    show_default=True,
+)
+@autocomplete_options
+def archaea(
+    ctx,
+    input,
+    output,
+    threads,
+    prefix,
+    evalue,
+    force,
+    autocomplete,
+    seed_value,
+    **kwargs,
+):
+    """Reorients your genome to begin with the archaeal COG1474 Orc1/cdc6 origin recognition complex gene"""
+
+    params = {
+        "--input": input,
+        "--output": output,
+        "--threads": threads,
+        "--force": force,
+        "--prefix": prefix,
+        "--evalue": evalue,
+        "--autocomplete": autocomplete,
+        "--seed_value": seed_value,
+        "--force": force,
+    }
+
+    # validates the directory  (need to before I start dnaapler or else no log file is written)
+    instantiate_dirs(output, force)
+
+    # defines gene
+    gene = "cog1474"
+
+    # initial logging etc
+    start_time = begin_dnaapler(input, output, threads, gene, params)
+    logger.info(
+        f"You have chosen {autocomplete} method to reorient your sequence if the BLAST based method fails."
+    )
+
+    # validates fasta
+    validate_fasta(input)
+
+    # validate e value
+    check_evalue(evalue)
+
+    # run BLAST
+    blast_success = run_blast_based_method(
+        ctx, input, output, prefix, gene, evalue, threads
+    )
+
+    # run autocomplete if BLAST reorientation failed
+    run_autocomplete(
+        blast_success, autocomplete, ctx, input, seed_value, output, prefix
+    )
+
+    # end dnaapler
+    end_dnaapler(start_time)
+
 """
 Plasmid command
 """

src/dnaapler/db/cog1474_db.pjs

@@ -0,0 +1,22 @@
+{
+  "version": "1.2",
+  "dbname": "cog1474_db",
+  "dbtype": "Protein",
+  "db-version": 5,
+  "description": "cog1474.fasta",
+  "number-of-letters": 158980,
+  "number-of-sequences": 403,
+  "last-updated": "2024-07-23T15:32:00",
+  "number-of-volumes": 1,
+  "bytes-total": 272156,
+  "bytes-to-cache": 162704,
+  "files": [
+    "cog1474_db.pdb",
+    "cog1474_db.phr",
+    "cog1474_db.pin",
+    "cog1474_db.pot",
+    "cog1474_db.psq",
+    "cog1474_db.ptf",
+    "cog1474_db.pto"
+  ]
+}

src/dnaapler/utils/cds_methods.py

@@ -205,7 +205,7 @@ def run_blast_based_method(
         input (Path): Path to the input DNA sequence file in FASTA format.
         output (Path): Path to the output directory where the reoriented sequence will be saved.
         prefix (str): Prefix for the output file name.
-        gene (str): Name of the gene used for BLAST search (options: 'dnaA', 'repA', 'terL', 'custom').
+        gene (str): Name of the gene used for BLAST search (options: 'dnaA', 'repA', 'terL', 'custom', 'cog1474').
         evalue (float): E-value threshold for BLAST search.
         threads (int): Number of threads for BLAST search.
 
@@ -223,6 +223,8 @@ def run_blast_based_method(
         db_name = "repA_db"
     elif gene == "terL":
         db_name = "terL_db"
+    elif gene == "cog1474":
+        db_name = "cog1474_db"
 
     # chromosome path
     # blast