Merge pull request #38 from RIVM-bioinformatics/bump_shigatyper

Bump shigatyper

Snakefile

@@ -42,7 +42,6 @@ include: "bin/rules/serotype.smk"
 include: "bin/rules/serotype_multireports.smk"
 include: "bin/rules/16s_extraction.smk"
 
-
 # @################################################################################
 # @####              Finalize pipeline (error/success)                        #####
 # @################################################################################
@@ -68,6 +67,7 @@ localrules:
     all,
     aggregate_serotypes,
     no_serotyper,
+    build_seroba_db,
 
 
 rule all:

bin/download_dbs.py

@@ -113,7 +113,7 @@ def copy_neisseria_db(self):
         ).exists():
             try:
                 print(
-                    f"Copying neisseria db from mnt/db/juno to: {characterize_neisseria_capsule_db_dir}"
+                    f"Copying neisseria db from /mnt/db/juno to: {characterize_neisseria_capsule_db_dir}"
                 )
                 build = subprocess.run(
                     [
@@ -181,7 +181,10 @@ def download_db_serotypefinder(self, version):
         return version
 
     def download_db_seroba(self, version, kmersize=71):
-        """Function to download the Seroba database if it is not present"""
+        """
+        Function to download the Seroba database if it is not present
+        Building is done in a Snakemake rule
+        """
         seroba_db_dir = self.db_dir.joinpath("seroba_db")
         if not seroba_db_dir.joinpath("database", "cdhit_cluster").is_file():
             print("\x1b[0;33m Downloading Seroba database...\n\033[0;0m")
@@ -203,14 +206,7 @@ def download_db_seroba(self, version, kmersize=71):
                     check=True,
                     timeout=60,
                 )
-                build = subprocess.run(
-                    ["seroba", "createDBs", "database", kmersize],
-                    cwd=str(seroba_db_dir),
-                    check=True,
-                    timeout=800,
-                )
             except (subprocess.CalledProcessError, subprocess.TimeoutExpired) as err:
-                build.kill()
                 raise err
         version = hf.get_commit_git(seroba_db_dir)
         return version

bin/rules/serotype.smk

@@ -141,10 +141,26 @@ rule ecoli_serotyper:
 ### Streptococcus pneumoniae serotyper ###
 
 
+rule build_seroba_db:
+    output:
+        temp(OUT + "/audit_trail/seroba_db_built.txt"),
+    conda:
+        "../../envs/seroba.yaml"
+    params:
+        seroba_db=config["seroba_db"],
+        kmer_size=config["seroba"]["kmer_size"],
+    shell:
+        """
+        cd {params.seroba_db}
+        seroba createDBs database {params.kmer_size}
+        """
+
+
 rule seroba:
     input:
         r1=lambda wildcards: SAMPLES[wildcards.sample]["R1"],
         r2=lambda wildcards: SAMPLES[wildcards.sample]["R2"],
+        check_db=OUT + "/audit_trail/seroba_db_built.txt",
     output:
         OUT + "/serotype/{sample}/pred.tsv",
     message:
@@ -199,7 +215,7 @@ rule shigatyper:
         CURRENT_DIR=$(pwd)
         cd "{params.output_dir}"
 
-        shigatyper {input.r1} {input.r2} > "$(basename {output.command_out})" 2> {log}
+        shigatyper --R1 {input.r1} --R2 {input.r2} | grep -E -A 1 '^sample' > "$(basename {output.command_out})" 2> {log}
 
         if [ -f {wildcards.sample}.csv ]
         then

envs/shigatyper.yaml

@@ -7,4 +7,4 @@ dependencies:
 - python=3.8.5
 - gsl=2.5
 - bcftools=1.9
-- shigatyper=1.0.7
+- shigatyper=2.0.5