Merge and upload reports

ktmeaton · Oct 28, 2020 · e1aa63b · e1aa63b
2 parents dfb3f0e + 7a99b75
commit e1aa63b
Show file tree

Hide file tree

Showing 5 changed files with 18 additions and 16 deletions.
diff --git a/profiles/compute-canada/config.yaml b/profiles/compute-canada/config.yaml
@@ -7,23 +7,24 @@ latency-wait: 5
 keep-going: True
 rerun-incomplete:  True
 printshellcmds: False
-use-singularity: True
-singularity-prefix: /scratch/keaton/singularity
-singularity-args: "-B /home -B /project -B /scratch -B /localscratch"
+#use-singularity: True
+#singularity-prefix: /scratch/keaton/singularity
+#singularity-args: "-B /home -B /project -B /scratch -B /localscratch"
 #log-handler-script: "workflow/scripts/slack_log.py"
 
 #------------------------------------------------------------------------------#
 # System config
 #------------------------------------------------------------------------------#
 # Total resources for the entire workflow
 # ex. 4 cpus, 4GB each (total 16GB), on one node
-resources: [cpus=4, mem_mb=16000, load=100]
+resources: [cpus=10, mem_mb=40000, load=100]
 # Default resources for a job
-default-resources: [cpus=1, mem_mb=2000, time_min=30, account="def-briang"]
+default-resources: [cpus=1, mem_mb=4000, time_min=60, account="def-briang"]
 # Maximum number of jobs to run
-jobs : "4"
+jobs : "10"
 # Where to put the .snakemake directory
 shadow-prefix: "/scratch/keaton/plague-phylogeography"
 # slurm execution
 cluster: "sbatch --parsable -t {resources.time_min} -A {resources.account} --cpus-per-task {resources.cpus} --mem-per-cpu={resources.mem_mb} -o workflow/logs_slurm/{rule}_%j.out -e workflow/logs_slurm/{rule}_%j.err --job-name {rule}"
 cluster-status: "workflow/scripts/slurm_status.py"
+scheduler: "greedy"
diff --git a/workflow/envs/main/environment.yaml b/workflow/envs/main/environment.yaml
@@ -8,7 +8,7 @@ dependencies:
   # Snakemake Execution
   - python=3.7.3
   - sqlite=3.31.1
-  - snakemake=5.25.0
+  - snakemake=5.26.1
   - flask=1.1.2
   - python-dotenv=0.14.0
   - slackclient=2.9.1

diff --git a/workflow/envs/merge/environment.yaml b/workflow/envs/merge/environment.yaml
@@ -66,7 +66,7 @@ dependencies:
 - samtools=1.10
 - singularity=3.6.3
 - slackclient=2.9.1
-- snakemake=5.25.0
+- snakemake=5.26.1
 - sqlite=3.31.1
 - vcftools=0.1.16
 - pip:

diff --git a/workflow/rules/download.smk b/workflow/rules/download.smk
@@ -10,7 +10,7 @@ rule download_sra:
   """
   Download SRA fastq files.
   """
-  message: "Downloading and dumping fastq files for BioSample {wildcards.sample}."
+  message: "Downloading and dumping {wildcards.file_acc} fastq for BioSample {wildcards.sample}."
   output:
     fastq = results_dir + "/data/sra/{sample}/{file_acc}_1.fastq.gz"
   log:

diff --git a/workflow/scripts/download_sra.sh b/workflow/scripts/download_sra.sh
@@ -21,14 +21,15 @@ SRA_ACC=$4
 #fi
 
 # Set cache enabled if not set
-if [[ -z `grep "/cache-enabled" $HOME/.ncbi/user-settings.mkfg` ]]; then
-  echo '/cache-enabled = "true"' >> $HOME/.ncbi/user-settings.mkfg
+if [[ -z `grep "cache-enabled" $HOME/.ncbi/user-settings.mkfg` ]]; then
+  vdb-config -s  "cache-enabled=true";
 fi;
 
 # Set the cache path
+
 if [[ -z `grep "/repository/user/main/public/root" $HOME/.ncbi/user-settings.mkfg` ]]; then\
   # Set SRA Cache Path
-  echo "/repository/user/main/public/root = "\"${CACHE_PATH}\" >> $HOME/.ncbi/user-settings.mkfg
+  vdb-config -s "/repository/user/main/public/root=${CACHE_PATH}";
 else
   # Retrieve SRA Cache Path
   CACHE_PATH=`grep "/repository/user/main/public/root" $HOME/.ncbi/user-settings.mkfg | \
@@ -38,12 +39,12 @@ fi;
 
 # Set the timeout
 if [[ -z `grep "/http/timeout/read" $HOME/.ncbi/user-settings.mkfg` ]]; then
-  echo '/http/timeout/read = "10000"' >> $HOME/.ncbi/user-settings.mkfg
+  vdb-config -s  "/http/timeout/read=10000";
 fi;
 
 # Echo for debugging
-#echo "SRA Cache:" ${CACHE_PATH}
-#echo "NCBI settings:" `cat $HOME/.ncbi/user-settings.mkfg`
+echo "SRA Cache:" ${CACHE_PATH}
+echo "NCBI settings:" `cat $HOME/.ncbi/user-settings.mkfg`
 
 #------------------------------------------------------------------------------#
 #                               SRA Download                                   #
@@ -66,7 +67,7 @@ if [[ ${validate_str} != *"corrupt"* ]]; then
     --outdir ${OUTDIR} \
     --skip-technical \
     --gzip \
-    --split-files ${SRA_ACC}; 
+    --split-files ${SRA_ACC};
   echo "SRA accession ${SRA_ACC} download completed for Biosample ${BIOSAMPLE_ACC}."
   mv ${OUTDIR}/${SRA_ACC}*.fastq.gz ${OUTDIR}/${BIOSAMPLE_ACC};
 else