updated rules for zymo download and prodigal runs

config/zymo/config.zymo.yaml

@@ -26,8 +26,8 @@ ref_genomes:
 
 
 # Additional params   
-single_fast5_dir: "data_single_fast5"
-multi_fast5_dir: "data_multi_fast5"
+# single_fast5_dir: "data_single_fast5"
+# multi_fast5_dir: "data_multi_fast5"
 
 # Annotation tools
 prodigal:

config/zymo/sbatch.zymo.sh

@@ -26,7 +26,7 @@ ONTP_SMK="workflow_zymo/Snakefile"
 # config file
 ONTP_CONFIG="config/zymo/config.zymo.yaml"
 # slurm config file
-ONTP_SLURM="config/Zymo/slurm.zymo.yaml"
+ONTP_SLURM="config/zymo/slurm.zymo.yaml"
 # slurm cluster call
 ONTP_CLUSTER="{cluster.call} -t {cluster.runtime} -n {cluster.threads} -p {cluster.partition} -N {cluster.nodes} -q {cluster.quality} --job-name={cluster.job-name}"
 

config/zymo/slurm.zymo.yaml

@@ -7,20 +7,20 @@ __default__:
   threads: 1 
   job-name: "ZYMO_REF.{rule}"
 
-download_genomes:
+zymo_download:
   partition: "interactive"
   quality: "normal"
   runtime: "0-00:15:00"
 
-concatenate:
+zymo_concat:
   partition: "batch"
   quality: "normal"
-  runtime: "0-00:05:00"
+  runtime: "0-00:10:00"
 
 zymo_prodigal:
   partition: "batch"
   qos: "normal"
-  runtime: "00-1:00:00"
+  runtime: "00-00:30:00"
 
 diamond_db:
   partition: "batch"
@@ -49,4 +49,4 @@ sibeliaz:
 zymo_sibeliaz:
   partition: "batch"
   quality: "normal"
-  runtime: "00-00:30:00"  
\ No newline at end of file
+  runtime: "00-00:30:00"  

workflow_zymo/Snakefile

@@ -27,8 +27,8 @@ workdir:
 # TARGETS
 ##################################################
 
-SINGLE_FAST5_DIR = os.path.abspath(config["single_fast5_dir"])
-MULTI_FAST5_DIR  = os.path.abspath(config["multi_fast5_dir"])
+# SINGLE_FAST5_DIR = os.path.abspath(config["single_fast5_dir"])
+# MULTI_FAST5_DIR  = os.path.abspath(config["multi_fast5_dir"])
 REF_GENOMES_DIR = os.path.abspath(config["ref_genomes_dir"])
 RESULTS_DIR = config["results_dir"]
 DATA_DIR = config["data_dir"]
@@ -54,11 +54,8 @@ READ_ASSEMBLER_PAIRS = assembler_pairs(READ_ASSEMBLERS)
 ##################################################
 # RULES
 ##################################################
-
 rule all:
     input:
-        REF_GENOMES_DIR,
-        expand(os.path.join(RESULTS_DIR,"reference/mock_genomes.fasta")),
         expand(os.path.join(RESULTS_DIR,"annotation/zymo_prodigal/proteins.faa")),
         expand(os.path.join(RESULTS_DIR,"annotation/diamond/zymo.dmnd")),
         expand(os.path.join(RESULTS_DIR, "analysis/diamond/{rtype_tool}.{ext}"), 
@@ -80,59 +77,50 @@ rule all:
             rtype_tool=["%s_%s" % (rtype, tool) for rtype, tool in READ_ASSEMBLERS])
 
 
-rule download_genomes:
+rule zymo_download:
     output:
-        directory(REF_GENOMES_DIR)
+        os.path.join(REF_GENOMES_DIR, "{wildcards.org}.fna")
     log:
-        out="logs/ref_genomes.out.log",
-        err="logs/ref_genomes.err.log"
-    shell:
-       "(date && "
-       "wget https://github.com/al-mcintyre/mCaller_analysis_scripts/raw/master/assemblies/bsubtilis_pb.fasta -P {output} && "
-       "wget https://github.com/al-mcintyre/mCaller_analysis_scripts/raw/master/assemblies/efaecalis_pb.fasta -P {output} && "
-       "wget https://github.com/al-mcintyre/mCaller_analysis_scripts/raw/master/assemblies/ecoli_pb.fasta -P {output} && "
-       "wget https://github.com/al-mcintyre/mCaller_analysis_scripts/raw/master/assemblies/lmonocytogenes_pb.fasta -P {output} && "
-       "wget https://github.com/al-mcintyre/mCaller_analysis_scripts/raw/master/assemblies/paeruginosa_pb.fasta -P {output} && "
-       "wget https://github.com/al-mcintyre/mCaller_analysis_scripts/raw/master/assemblies/scerevisiae_pb.fasta -P {output} && "
-       "wget https://github.com/al-mcintyre/mCaller_analysis_scripts/raw/master/assemblies/senterica_pb.fasta -P {output} && "
-       "wget https://github.com/al-mcintyre/mCaller_analysis_scripts/raw/master/assemblies/saureus_pb.fasta -P {output} && "
-       "wget https://github.com/al-mcintyre/mCaller_analysis_scripts/raw/master/assemblies/lfermentum_ontlumina.fasta -P {output} && "
-       "wget https://github.com/al-mcintyre/mCaller_analysis_scripts/raw/master/assemblies/cneoformans_ont.fasta -P {output} && "
-       "date) 2> {log.err} > {log.out}"
-
-rule concatenate:
-    input:
-        rules.download_genomes.output
-    output:
-        os.path.join(RESULTS_DIR,"reference/mock_genomes.fasta")
-    log:
-        out="logs/concat.out.log",
-        err="logs/concat.err.log"
+        "logs/zymo.download.{wildcards.org}.log"
+    params:
+        url=lambda wildcards: config["ref_genomes"][wildcards.org]
     shell:
-        "(date && "
-        "cat {input}/*.fasta > {output} && "
-        "date) 2> {log.err} > {log.out}"
+       "(date && wget {params.url} -P {output} && date) &> {log}"
 
 rule zymo_prodigal:
     input:
-        rules.concatenate.output
+        rules.zymo_download.output
     output:
-        os.path.join(RESULTS_DIR,"annotation/zymo_prodigal/proteins.faa")
+        faa=os.path.join(RESULTS_DIR,"annotation/zymo_prodigal/{wildcards.org}.faa"),
+        gbk=os.path.join(RESULTS_DIR,"annotation/zymo_prodigal/{wildcards.org}.gbk")
     log:
-        out="logs/zymo_prodigal.out.log",
-        err="logs/zymo_prodigal.err.log"
+        out="logs/zymo_prodigal.{wildcards.org}.out.log",
+        err="logs/zymo_prodigal.{wildcards.org}.err.log"
     threads:
         config["prodigal"]["threads"]
     conda:
         os.path.join(ENV_DIR, "prodigal.yaml")
     shell:
        "(date && "
-       "prodigal -a {output} -p meta -i {input} && "
+       "prodigal -i {input} -o {output.gbk} -a {output.faa} && "
+       "sed -i 's/^>/>{wildcards.org}__/' {output.faa} && "
        "date) 2> {log.err} > {log.out}"
 
+rule zymo_concat:
+    input:
+        expand(
+            os.path.join(RESULTS_DIR, "annotation/zymo_prodigal/{org}.faa"),
+            org=config["ref_genomes"].keys()
+        )
+    output:
+        os.path.join(RESULTS_DIR, "annotation/zymo_prodigal/proteins.faa")
+    shell:
+        "cat {input} > {output}"
+
+
 rule diamond_db:
     input:
-        rules.zymo_prodigal.output
+        rules.zymo_concat.output
     output:
         os.path.join(RESULTS_DIR,"annotation/diamond/zymo.dmnd")
     log:
@@ -285,6 +273,19 @@ rule sibeliaz:
        "sibeliaz -n -t {threads} -k {params.kmers} -f {params.memory} -o $(dirname {output}) {input.asm1} {input.asm2} && "
        "date) 2> {log.err} > {log.out}"
 
+rule concatenate:
+    input:
+        rules.zymo_download.output
+    output:
+        os.path.join(RESULTS_DIR,"reference/mock_genomes.fasta")
+    log:
+        out="logs/concat.out.log",
+        err="logs/concat.err.log"
+    shell:
+        "(date && "
+        "cat {input}/*.fasta > {output} && "
+        "date) 2> {log.err} > {log.out}"
+
 rule zymo_sibeliaz:
     input:
         zymo=rules.concatenate.output,