From c8c9a1e8c500ba7b7c52950b561c407cf272fc57 Mon Sep 17 00:00:00 2001
From: Kelly Sovacool <kelly.sovacool@nih.gov>
Date: Mon, 29 Apr 2024 17:42:20 -0400
Subject: [PATCH] feat: create circRNA_finder docker

---
 config/config.yaml                    |  7 ++++--
 docker/circRNA_finder/Dockerfile      | 32 +++++++++++++++++++++++++++
 docker/circRNA_finder/environment.txt |  2 ++
 docker/circRNA_finder/meta.yml        |  4 ++++
 workflow/rules/findcircrna.smk        |  7 ++----
 5 files changed, 45 insertions(+), 7 deletions(-)
 create mode 100644 docker/circRNA_finder/Dockerfile
 create mode 100644 docker/circRNA_finder/environment.txt
 create mode 100644 docker/circRNA_finder/meta.yml

diff --git a/config/config.yaml b/config/config.yaml
index 8040726..d66a139 100644
--- a/config/config.yaml
+++ b/config/config.yaml
@@ -104,9 +104,11 @@ cutadapt_q: 20
 high_confidence_core_callers: "circExplorer,circExplorer_bwa"
 high_confidence_core_callers_plus_n: 1
 
-ciri_perl_script: "/opt2/CIRI_v2.0.6/CIRI2.pl" # in docker container
-circrnafinder_dir: "/data/CCBR_Pipeliner/bin/circRNA_finder-1.2"
+# paths in docker containers
+ciri_perl_script: "/opt2/CIRI_v2.0.6/CIRI2.pl" 
 find_circ_dir: "/data/CCBR_Pipeliner/bin/find_circ"
+
+# change this path to a directory containing fasta and GTF files for all host and virus genomes
 fastas_gtfs_dir: "/data/CCBR_Pipeliner/db/PipeDB/charlie/fastas_gtfs"
 
 annotation_lookups:
@@ -117,6 +119,7 @@ containers:
   base: "docker://nciccbr/ccbr_ubuntu_base_20.04:v6"
   bowtie1: "docker://nciccbr/charlie_bowtie1:v0.1.0"
   circexplorer: 'docker://nciccbr/ccbr_circexplorer:v1.0'
+  circRNA_finder: 'docker://nciccbr/charlie_circRNA_finder:v1'
   ciri: 'docker://nciccbr/charlie_ciri2:v1'
   clear: "docker://nciccbr/ccbr_clear:latest"
   cutadapt: 'docker://nciccbr/charlie_cutadapt_fqfilter:v1'
diff --git a/docker/circRNA_finder/Dockerfile b/docker/circRNA_finder/Dockerfile
new file mode 100644
index 0000000..f119e9a
--- /dev/null
+++ b/docker/circRNA_finder/Dockerfile
@@ -0,0 +1,32 @@
+FROM nciccbr/ccbr_ubuntu_base_20.04:v6
+
+# build time variables
+ARG BUILD_DATE="000000"
+ENV BUILD_DATE=${BUILD_DATE}
+ARG BUILD_TAG="000000"
+ENV BUILD_TAG=${BUILD_TAG}
+ARG REPONAME="000000"
+ENV REPONAME=${REPONAME}
+
+# install conda packages
+COPY environment.txt /data2/
+RUN mamba install -c conda-forge -c bioconda --file /data2/environment.txt
+ENV R_LIBS_USER=/opt2/conda/lib/R/library/
+
+# install circRNA_finder
+WORKDIR /opt2
+ENV VERSION=v1.2
+RUN wget https://github.com/orzechoj/circRNA_finder/archive/refs/tags/v${VERSION}.tar.gz -O circRNA.tar.gz && \
+    tar -xzvf circRNA.tar.gz && \
+    cd circRNA_finder-${VERSION} && \
+    python setup.py install
+ENV PATH="/opt2/circRNA_finder-${VERSION}/:$PATH"
+
+# Save Dockerfile in the docker
+COPY Dockerfile /opt2/Dockerfile_${REPONAME}.${BUILD_TAG}
+RUN chmod a+r /opt2/Dockerfile_${REPONAME}.${BUILD_TAG}
+
+# cleanup
+WORKDIR /data2
+RUN apt-get clean && apt-get purge \
+    && rm -rf /var/lib/apt/lists/* /tmp/* /var/tmp/*
diff --git a/docker/circRNA_finder/environment.txt b/docker/circRNA_finder/environment.txt
new file mode 100644
index 0000000..fd233e3
--- /dev/null
+++ b/docker/circRNA_finder/environment.txt
@@ -0,0 +1,2 @@
+samtools
+STAR
\ No newline at end of file
diff --git a/docker/circRNA_finder/meta.yml b/docker/circRNA_finder/meta.yml
new file mode 100644
index 0000000..2f1cc4c
--- /dev/null
+++ b/docker/circRNA_finder/meta.yml
@@ -0,0 +1,4 @@
+dockerhub_namespace: nciccbr
+image_name: charlie_circRNA_finder
+version: v1
+container: "$(dockerhub_namespace)/$(image_name):$(version)"
diff --git a/workflow/rules/findcircrna.smk b/workflow/rules/findcircrna.smk
index e08d3a6..46eaca9 100644
--- a/workflow/rules/findcircrna.smk
+++ b/workflow/rules/findcircrna.smk
@@ -1123,12 +1123,9 @@ rule circrnafinder:
             "{sample}.Chimeric.out.sorted.bam",
         ),
     params:
-        postProcessStarAlignment_script=join(
-            config["circrnafinder_dir"], "postProcessStarAlignment.pl"
-        ),
         bsj_min_nreads=config["minreadcount"],
         randomstr=str(uuid.uuid4()),
-    container: config['containers']['base']
+    container: config['containers']['circRNA_finder']
     shell:
         """
 set -exo pipefail
@@ -1145,7 +1142,7 @@ outDir=$(dirname {output.bed})
 if [ -d $outDir ];then rm -rf $outDir;fi
 if [ ! -d $outDir ];then mkdir -p $outDir;fi
 
-{params.postProcessStarAlignment_script} \\
+postProcessStarAlignment.pl \\
     --starDir ${{starDir}}/ \\
     --outDir ${{outDir}}/