From 37d6306530cccbb61d503e070fb50b8e85405f56 Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?Sameer=20D=E2=80=99Costa?= <sameerdcosta@gmail.com>
Date: Fri, 15 Feb 2019 11:49:40 -0600
Subject: [PATCH] Changes to get parliament2 working with Singularity on HPC
 systems (#46)

* Adding Singularity file
spython recipe DockerFile > Singularity

* Trying to create resources.tar.gz file beforehand

* Changing tar options

* tar xtract

* try different version of chmod

* Setting entrypoint

* Moving back to runscript

* Fixing paths for singularity

* Minor change to maintainer flag

* Fix manta path

* need to learn how to trigger a singularity build.

* Changes to paths for survivor and svtyper

* trying to run svviz

* typo in svviz

* trigger

* Removing memfree option from parallel

* Conda environment for svtyper

* removing double sourcing of conda.sh

* Creating environment for svtyper

* changing dependency to python 2.7

* making Singularity changes to Dockerfile

* build hook to create resources.tar.gz on dockerhub

* Removing Singularity build file as now we are able to build with docker
and run with singularity

* Smaller travis test at the end

* Changing test to just do Breakdancer and SVTyper
---
 .travis.yml                                   |   5 +-
 Dockerfile                                    |   6 +
 hooks/pre_build                               |   5 +
 parliament2.sh                                | 106 ++++++++++--------
 .../home/dnanexus/parallelize_svtyper.sh      |   6 +-
 resources/usr/bin/runManta                    |   2 +-
 svtyper_env.yml                               |   6 +
 7 files changed, 84 insertions(+), 52 deletions(-)
 create mode 100644 hooks/pre_build
 create mode 100644 svtyper_env.yml

diff --git a/.travis.yml b/.travis.yml
index 7e3e10d5..6e6db4b8 100644
--- a/.travis.yml
+++ b/.travis.yml
@@ -28,7 +28,8 @@ jobs:
     - script: docker run -v /home/dnanexus/in:/home/dnanexus/in -v /home/dnanexus/out:/home/dnanexus/out dnanexus/parliament2:$TAG --bam /home/dnanexus/in/small_input.bam --bai /home/dnanexus/in/small_input.bai --ref_genome /home/dnanexus/in/ref.fa.gz --fai /home/dnanexus/in/ref.fa.fai --prefix lumpy --lumpy && ls -sh /home/dnanexus/out
     - script: docker run -v /home/dnanexus/in:/home/dnanexus/in -v /home/dnanexus/out:/home/dnanexus/out dnanexus/parliament2:$TAG --bam /home/dnanexus/in/small_input.bam --bai /home/dnanexus/in/small_input.bai --ref_genome /home/dnanexus/in/ref.fa.gz --fai /home/dnanexus/in/ref.fa.fai --prefix manta --manta && ls -sh /home/dnanexus/out
     - script: docker run -v /home/dnanexus/in:/home/dnanexus/in -v /home/dnanexus/out:/home/dnanexus/out dnanexus/parliament2:$TAG --bam /home/dnanexus/in/small_input.bam --bai /home/dnanexus/in/small_input.bai --ref_genome /home/dnanexus/in/ref.fa.gz --fai /home/dnanexus/in/ref.fa.fai --prefix svviz --breakdancer --svviz && ls -sh /home/dnanexus/out
-    - script: docker run -v /home/dnanexus/in:/home/dnanexus/in -v /home/dnanexus/out:/home/dnanexus/out dnanexus/parliament2:$TAG --bam /home/dnanexus/in/small_input.bam --bai /home/dnanexus/in/small_input.bai --ref_genome /home/dnanexus/in/ref.fa.gz --fai /home/dnanexus/in/ref.fa.fai --prefix full --breakdancer --breakseq --cnvnator --delly_deletion --delly_duplication --delly_insertion --delly_inversion --lumpy --manta && ls -sh /home/dnanexus/out
+    - script: docker run -v /home/dnanexus/in:/home/dnanexus/in -v /home/dnanexus/out:/home/dnanexus/out dnanexus/parliament2:$TAG --bam /home/dnanexus/in/small_input.bam --bai /home/dnanexus/in/small_input.bai --ref_genome /home/dnanexus/in/ref.fa.gz --fai /home/dnanexus/in/ref.fa.fai --prefix full --breakdancer --genotype && ls -sh /home/dnanexus/out
+
 
 deploy:
   provider: script
@@ -42,4 +43,4 @@ branches:
 
 language: python
 python: 
-  - "2.7.13"
\ No newline at end of file
+  - "2.7.13"
diff --git a/Dockerfile b/Dockerfile
index 94f7e771..6daa7346 100644
--- a/Dockerfile
+++ b/Dockerfile
@@ -96,6 +96,12 @@ RUN mkdir -p /home/dnanexus/in /home/dnanexus/out
 WORKDIR /home/dnanexus
 COPY parliament2.py .
 COPY parliament2.sh .
+COPY svtyper_env.yml .
+
+RUN conda create -y --name svviz_env svviz
+# We have to use a slightly different method for 
+# svtyper as it installs software directly from git 
+RUN conda env create --name svtyper_env --file svtyper_env.yml
 
 RUN /bin/bash -c "source /etc/profile.d/dnanexus.environment.sh"
 
diff --git a/hooks/pre_build b/hooks/pre_build
new file mode 100644
index 00000000..1644d08f
--- /dev/null
+++ b/hooks/pre_build
@@ -0,0 +1,5 @@
+#!/bin/bash
+
+echo "Compressing the resources directory"
+
+tar -czf resources.tar.gz resources/
diff --git a/parliament2.sh b/parliament2.sh
index 5e8952ef..5084d7f3 100644
--- a/parliament2.sh
+++ b/parliament2.sh
@@ -75,11 +75,11 @@ fi
 ref_genome=$(python /home/dnanexus/get_reference.py)
 lumpy_exclude_string=""
 if [[ "${ref_genome}" == "b37" ]]; then
-    lumpy_exclude_string="-x b37.bed"
+    lumpy_exclude_string="-x /home/dnanexus/b37.bed"
 elif [[ "$ref_genome" == "hg19" ]]; then
-    lumpy_exclude_string="-x hg19.bed"
+    lumpy_exclude_string="-x /home/dnanexus/hg19.bed"
 else
-    lumpy_exclude_string="-x hg38.bed"
+    lumpy_exclude_string="-x /home/dnanexus/hg38.bed"
 fi
 
 export lumpy_scripts="/home/dnanexus/lumpy-sv/scripts"
@@ -129,8 +129,8 @@ else
     touch /home/dnanexus/in/done.txt
 fi
 
-ln -s /home/dnanexus/in/input.bam /home/dnanexus/input.bam
-ln -s /home/dnanexus/in/input.bam.bai /home/dnanexus/input.bam.bai
+ln -s /home/dnanexus/in/input.bam
+ln -s /home/dnanexus/in/input.bam.bai
 
 wait
 
@@ -151,7 +151,7 @@ if [[ "${run_breakseq}" == "True" ]]; then
     mkdir -p /home/dnanexus/out/log_files/breakseq_logs/
     bplib="/breakseq2_bplib_20150129/breakseq2_bplib_20150129.gff"
     work="breakseq2"
-    timeout 6h ./breakseq2-2.2/scripts/run_breakseq2.py --reference ref.fa \
+    timeout 6h /home/dnanexus/breakseq2-2.2/scripts/run_breakseq2.py --reference ref.fa \
         --bams input.bam --work "${work}" \
         --bwa /usr/local/bin/bwa --samtools /usr/local/bin/samtools \
         --bplib_gff "${bplib}" \
@@ -261,7 +261,7 @@ if [[ "${run_cnvnator}" == "True" ]] || [[ "${run_delly}" == "True" ]] || [[ "${
 
                 if [[ "${run_lumpy}" == "True" ]]; then
                     echo "Running Lumpy for contig ${contig}"
-                    timeout 6h ./lumpy-sv/bin/lumpyexpress -B chr."${count}".bam -o lumpy."${count}".vcf ${lumpy_exclude_string} -k 1> /home/dnanexus/out/log_files/lumpy_logs/"${prefix}".lumpy."${count}".stdout.log 2> /home/dnanexus/out/log_files/lumpy_logs/"${prefix}".lumpy."${count}".stderr.log & 
+                    timeout 6h /home/dnanexus/lumpy-sv/bin/lumpyexpress -B chr."${count}".bam -o lumpy."${count}".vcf ${lumpy_exclude_string} -k 1> /home/dnanexus/out/log_files/lumpy_logs/"${prefix}".lumpy."${count}".stdout.log 2> /home/dnanexus/out/log_files/lumpy_logs/"${prefix}".lumpy."${count}".stderr.log & 
                     lumpy_merge_command="$lumpy_merge_command lumpy.$count.vcf"
                 fi
             fi
@@ -274,11 +274,6 @@ fi
 
 wait
 
-# Only install SVTyper if necessary
-if [[ "${run_genotype_candidates}" == "True" ]]; then
-    pip install git+https://github.com/hall-lab/svtyper.git -q &
-fi
-
 echo "Converting results to VCF format"
 mkdir -p /home/dnanexus/out/sv_caller_results/
 
@@ -346,6 +341,19 @@ fi) &
 
 (if [[ "${run_breakseq}" == "True" ]]; then
     echo "Convert Breakseq results to VCF format"
+    if [[ ! -f breakseq2/breakseq_genotyped.gff && ! -f breakseq2/breakseq.vcf.gz && ! -f breakseq2/final.bam ]]; then
+        echo "No outputs of Breakseq found. Continuing."
+    else
+        mv breakseq2/breakseq.vcf.gz .
+        gunzip breakseq.vcf.gz
+
+        cp breakseq2/breakseq_genotyped.gff /home/dnanexus/out/sv_caller_results/"${prefix}".breakseq.gff
+        cp breakseq.vcf /home/dnanexus/out/sv_caller_results/"${prefix}".breakseq.vcf
+        cp breakseq2/final.bam /home/dnanexus/out/sv_caller_results/"${prefix}".breakseq.bam
+    fi
+
+    # Do the log files after we copy the output so that the 
+    # cd /home/dnanexus command doesn't spoil singularity
     if [[ -z $(find "${work}" -name "*.log") ]]; then
         echo "No Breakseq log files found."
     else
@@ -356,16 +364,7 @@ fi) &
         cd /home/dnanexus || return
     fi
 
-    if [[ ! -f breakseq2/breakseq_genotyped.gff && ! -f breakseq2/breakseq.vcf.gz && ! -f breakseq2/final.bam ]]; then
-        echo "No outputs of Breakseq found. Continuing."
-    else
-        mv breakseq2/breakseq.vcf.gz .
-        gunzip breakseq.vcf.gz
 
-        cp breakseq2/breakseq_genotyped.gff /home/dnanexus/out/sv_caller_results/"${prefix}".breakseq.gff
-        cp breakseq.vcf /home/dnanexus/out/sv_caller_results/"${prefix}".breakseq.vcf
-        cp breakseq2/final.bam /home/dnanexus/out/sv_caller_results/"${prefix}".breakseq.bam
-    fi
 fi) &
 
 (if [[ "${run_delly_deletion}" == "True" ]]; then 
@@ -428,24 +427,25 @@ set +e
 
 # Run SVtyper and SVviz
 if [[ "${run_genotype_candidates}" == "True" ]]; then
-    echo "Running SVTyper"
-    # SVviz and BreakSeq have mutually exclusive versions of pysam required, so
-    # SVviz is only installed later and if necessary
-    if [[ "${run_svviz}" == "True" ]]; then
-        pip install svviz -q &
-    fi
 
+    # Only install SVTyper if necessary
+    #pip install git+https://github.com/hall-lab/svtyper.git -q &
+    source /miniconda/etc/profile.d/conda.sh
+    conda activate svtyper_env
+
+
+    echo "Running SVTyper"
     mkdir -p /home/dnanexus/out/svtyped_vcfs/
 
     i=0
     # Breakdancer
     if [[ "${run_breakdancer}" == "True" ]]; then
         echo "Running SVTyper on Breakdancer outputs"
-        mkdir /home/dnanexus/svtype_breakdancer
-        if [[ -f /home/dnanexus/breakdancer.vcf ]]; then
-            bash ./parallelize_svtyper.sh /home/dnanexus/breakdancer.vcf svtype_breakdancer /home/dnanexus/"${prefix}".breakdancer.svtyped.vcf input.bam
+        mkdir svtype_breakdancer
+        if [[ -f breakdancer.vcf ]]; then
+            bash /home/dnanexus/parallelize_svtyper.sh breakdancer.vcf svtype_breakdancer "${prefix}".breakdancer.svtyped.vcf input.bam
 
-            sed -i 's/SAMPLE/breakdancer/g' /home/dnanexus/"${prefix}".breakdancer.svtyped.vcf
+            sed -i 's/SAMPLE/breakdancer/g' "${prefix}".breakdancer.svtyped.vcf
         else
             "No Breakdancer VCF file found. Continuing."
         fi
@@ -454,9 +454,9 @@ if [[ "${run_genotype_candidates}" == "True" ]]; then
     # Breakseq
     if [[ "${run_breakseq}" == "True" ]]; then
         echo "Running SVTyper on BreakSeq outputs"
-        mkdir /home/dnanexus/svtype_breakseq
-        if [[ -f /home/dnanexus/breakseq.vcf ]]; then
-            bash ./parallelize_svtyper.sh /home/dnanexus/breakseq.vcf svtype_breakseq /home/dnanexus/"${prefix}".breakseq.svtyped.vcf input.bam
+        mkdir svtype_breakseq
+        if [[ -f breakseq.vcf ]]; then
+            bash /home/dnanexus/parallelize_svtyper.sh breakseq.vcf svtype_breakseq "${prefix}".breakseq.svtyped.vcf input.bam
         else
             echo "No BreakSeq VCF file found. Continuing."
         fi
@@ -465,10 +465,10 @@ if [[ "${run_genotype_candidates}" == "True" ]]; then
     # CNVnator
     if [[ "${run_cnvnator}" == "True" ]]; then
         echo "Running SVTyper on CNVnator outputs"
-        mkdir /home/dnanexus/svtype_cnvnator
-        if [[ -f /home/dnanexus/cnvnator.vcf ]]; then
-            python /get_uncalled_cnvnator.py | python /add_ciend.py 1000 > /home/dnanexus/cnvnator.ci.vcf < cnvnator.vcf
-            bash ./parallelize_svtyper.sh /home/dnanexus/cnvnator.vcf svtype_cnvnator "${prefix}".cnvnator.svtyped.vcf input.bam
+        mkdir svtype_cnvnator
+        if [[ -f cnvnator.vcf ]]; then
+            python /get_uncalled_cnvnator.py | python /add_ciend.py 1000 > cnvnator.ci.vcf < cnvnator.vcf
+            bash /home/dnanexus/parallelize_svtyper.sh cnvnator.vcf svtype_cnvnator "${prefix}".cnvnator.svtyped.vcf input.bam
         else
             echo "No CNVnator VCF file found. Continuing."
         fi
@@ -481,8 +481,8 @@ if [[ "${run_genotype_candidates}" == "True" ]]; then
             echo "No Delly VCF file found. Continuing."
         else
             for item in delly*vcf; do
-                mkdir /home/dnanexus/svtype_delly_"${i}"
-                bash ./parallelize_svtyper.sh /home/dnanexus/"${item}" svtype_delly_"${i}" /home/dnanexus/delly.svtyper."${i}".vcf input.bam
+                mkdir svtype_delly_"${i}"
+                bash /home/dnanexus/parallelize_svtyper.sh "${item}" svtype_delly_"${i}" delly.svtyper."${i}".vcf input.bam
                 i=$((i + 1))
             done
 
@@ -497,9 +497,9 @@ if [[ "${run_genotype_candidates}" == "True" ]]; then
     # Lumpy
     if [[ "${run_lumpy}" == "True" ]]; then
         echo "Running SVTyper on Lumpy outputs"
-        mkdir /home/dnanexus/svtype_lumpy
-        if [[ -f /home/dnanexus/lumpy.vcf ]]; then
-            bash ./parallelize_svtyper.sh /home/dnanexus/lumpy.vcf svtype_lumpy /home/dnanexus/"${prefix}".lumpy.svtyped.vcf input.bam
+        mkdir svtype_lumpy
+        if [[ -f lumpy.vcf ]]; then
+            bash /home/dnanexus/parallelize_svtyper.sh lumpy.vcf svtype_lumpy "${prefix}".lumpy.svtyped.vcf input.bam
         else
             echo "No Lumpy VCF file found. Continuing."
         fi
@@ -509,7 +509,7 @@ if [[ "${run_genotype_candidates}" == "True" ]]; then
     if [[ "${run_manta}" == "True" ]]; then
         echo "Running SVTyper on Manta outputs"
         if [[ -f diploidSV.vcf ]]; then
-            mv diploidSV.vcf /home/dnanexus/"${prefix}".manta.svtyped.vcf
+            mv diploidSV.vcf "${prefix}".manta.svtyped.vcf
         else
             echo "No Manta VCF file found. Continuing."
         fi
@@ -517,6 +517,9 @@ if [[ "${run_genotype_candidates}" == "True" ]]; then
 
     wait
 
+    # deactivate svtyper
+    source deactivate
+
     # Prepare inputs for SURVIVOR
     echo "Preparing inputs for SURVIVOR"
     for item in *svtyped.vcf; do
@@ -543,9 +546,16 @@ if [[ "${run_genotype_candidates}" == "True" ]]; then
 
     # Run svviz
     if [[ "${run_svviz}" == "True" ]]; then
+
+        # SVviz and BreakSeq have mutually exclusive versions of pysam required, so
+        # SVviz is only installed later and if necessary
+        conda activate svviz_env
+        #pip install svviz -q &
+
+
         echo "Running svviz"
         mkdir -p /home/dnanexus/out/log_files/svviz_logs/
-        mkdir /home/dnanexus/svviz_outputs
+        mkdir svviz_outputs
 
         grep \# survivor_sorted.vcf > header.txt
 
@@ -576,9 +586,11 @@ if [[ "${run_genotype_candidates}" == "True" ]]; then
             
             threads="$(nproc)"
             threads=$((threads / 2))
-            parallel --memfree 5G --retries 2 --verbose -a commands.txt eval 1>/home/dnanexus/out/log_files/svviz_logs/svviz.stdout.log 2>/home/dnanexus/out/log_files/svviz_logs/svviz.stderr.log
+            # removing the memfree option as it doesn't seem to exist in Ubuntu 14.04
+            #parallel --memfree 5G --retries 2 --verbose -a commands.txt eval 1>/home/dnanexus/out/log_files/svviz_logs/svviz.stdout.log 2>/home/dnanexus/out/log_files/svviz_logs/svviz.stderr.log
+            parallel --retries 2 --verbose -a commands.txt eval 1>/home/dnanexus/out/log_files/svviz_logs/svviz.stdout.log 2>/home/dnanexus/out/log_files/svviz_logs/svviz.stderr.log
             
-            cd /home/dnanexus/svviz_outputs && tar -czf /home/dnanexus/out/"${prefix}".svviz_outputs.tar.gz .
+            cd svviz_outputs && tar -czf /home/dnanexus/out/"${prefix}".svviz_outputs.tar.gz .
         fi
     fi
 fi
diff --git a/resources/home/dnanexus/parallelize_svtyper.sh b/resources/home/dnanexus/parallelize_svtyper.sh
index 8bbd7296..cd148584 100755
--- a/resources/home/dnanexus/parallelize_svtyper.sh
+++ b/resources/home/dnanexus/parallelize_svtyper.sh
@@ -21,9 +21,11 @@ for item in $directory*; do
     echo "svtyper -B $input_bam -i $directory/$i >> $directory/$i" >> $output.cmds
 done
 
-parallel --memfree 5G --retries 2 --verbose -a $output.cmds eval 2> /dev/null
+# We don't have the memfree option is the Ubuntu 14.04 version  of parallel
+#parallel --memfree 5G --retries 2 --verbose -a $output.cmds eval 2> /dev/null
+parallel --retries 2 --verbose -a $output.cmds eval 2> /dev/null
 
 grep \# $input > $output
 for item in $directory/*; do
     grep -v \# $item >> $output
-done
\ No newline at end of file
+done
diff --git a/resources/usr/bin/runManta b/resources/usr/bin/runManta
index ad0e2cae..8f4aeae0 100755
--- a/resources/usr/bin/runManta
+++ b/resources/usr/bin/runManta
@@ -7,4 +7,4 @@ done < contigs
 
 python /miniconda/bin/configManta.py --referenceFasta ref.fa --normalBam input.bam --runDir manta $region_string
 
-python /home/dnanexus/manta/runWorkflow.py -m local -j 16 
\ No newline at end of file
+python ./manta/runWorkflow.py -m local -j 16 
diff --git a/svtyper_env.yml b/svtyper_env.yml
new file mode 100644
index 00000000..35b2bcc2
--- /dev/null
+++ b/svtyper_env.yml
@@ -0,0 +1,6 @@
+name: svtyper_env
+channels:
+dependencies:
+   - python=2.7
+   - pip:
+     - "--editable=git+https://github.com/hall-lab/svtyper.git#egg=svtyper_git"