Merge pull request #19 from sjaenick/main

umi_tools reproducibility issues

PARANOiD.nf

@@ -358,7 +358,7 @@ process deduplicate{
 	file "${query.baseName}.deduplicated.log*" into log_deduplicate_to_collect_statistics
 
 	"""
-	umi_tools dedup -I ${query} --output-stats ${query.baseName}.deduplicated.log -S ${query.baseName}.deduplicated.bam
+	umi_tools dedup --random-seed=42 -I ${query} --output-stats ${query.baseName}.deduplicated.log -S ${query.baseName}.deduplicated.bam
 	"""
 }
 

dockerfiles/dockerfile.umi-tools

@@ -6,8 +6,16 @@ LABEL tool.version="1.1.2"
 MAINTAINER Patrick Barth <[email protected]>
 
 RUN apt-get update -y && \
-    apt-get install -y gcc procps && \
-    pip install umi-tools==1.1.2 && \
-    apt-get purge -y gcc && \
+    apt-get install -y gcc procps patch wget && \
+    wget -q https://github.com/CGATOxford/UMI-tools/archive/refs/tags/1.1.2.tar.gz && \
+    tar xzf 1.1.2.tar.gz && \
+    cd UMI-tools-1.1.2 && \
+    wget -q -O - https://patch-diff.githubusercontent.com/raw/CGATOxford/UMI-tools/pull/550.diff | patch -p1 && \
+    python setup.py install && \
+    cd .. && \
+    rm -rf UMI-tools-1.1.2 1.1.2.tar.gz && \
+    apt-get purge -y gcc patch wget && \
     apt-get -y autoremove
 
+CMD ["umi_tools"]
+