Better matches EDGE output

LANL-Bioinformatics · Aug 7, 2024 · 3645606 · 3645606
1 parent 1a51cb3
commit 3645606
Show file tree

Hide file tree

Showing 4 changed files with 137 additions and 75 deletions.
diff --git a/runAssembly/Dockerfile b/runAssembly/Dockerfile
@@ -1,32 +1,54 @@
-FROM continuumio/miniconda3:23.5.2-0
+FROM continuumio/miniconda3:23.5.2-0 AS build
 
 ENV container=docker
 
 # add conda channels
 RUN conda config --add channels conda-forge \
     && conda config --add channels bioconda
 
+RUN conda init bash \
+    && . ~/.bashrc \
+    && conda create --name assembly \
+    && conda activate assembly 
+
 # install dependencies
 RUN conda install conda-libmamba-solver
 RUN conda config --set solver libmamba
-RUN conda install -c conda-forge python=3.9
-RUN conda install -c bioconda samtools=1.17
-RUN conda install -c bioconda racon=1.5
-RUN conda install -c bioconda seqtk=1.3
-RUN conda install -c bioconda spades=3.15.5
-RUN conda install -c bioconda minimap2=2.26
-RUN conda install -c bioconda megahit=1.2.9
-RUN conda install -c bioconda idba=1.1.3
-RUN conda install -c bioconda unicycler=0.5.0
-RUN wget https://github.com/ruanjue/wtdbg2/releases/download/v2.5/wtdbg-2.5_x64_linux.tgz
-RUN tar -xvzf wtdbg-2.5_x64_linux.tgz
-RUN cp wtdbg-2.5_x64_linux/* /opt/conda/bin
-RUN conda install -c bioconda flye=2.9.2
+RUN conda install -n assembly -c conda-forge python=3.9
+RUN conda install -n assembly -c bioconda samtools=1.17
+RUN conda install -n assembly -c bioconda racon=1.5
+RUN conda install -n assembly -c bioconda seqtk=1.3
+RUN conda install -n assembly -c bioconda spades=3.15.5
+RUN conda install -n assembly -c bioconda minimap2=2.26
+RUN conda install -n assembly -c bioconda megahit=1.2.9
+RUN conda install -n assembly -c bioconda idba=1.1.3
+RUN conda install -n assembly -c bioconda unicycler=0.5.0
+RUN wget https://github.com/ruanjue/wtdbg2/releases/download/v2.5/wtdbg-2.5_x64_linux.tgz \
+    && tar -xvzf wtdbg-2.5_x64_linux.tgz \
+    && cp wtdbg-2.5_x64_linux/* /opt/conda/envs/assembly/bin
+RUN conda install -n assembly -c bioconda flye=2.9.2
+RUN conda install -n assembly git
+RUN conda install -c conda-forge conda-pack
+
+ADD bin/extractLongReads.pl /opt/conda/envs/assembly/bin
+ADD bin/getAvgLen.pl /opt/conda/envs/assembly/bin
+ADD bin/renameFilterFasta.pl /opt/conda/envs/assembly/bin
+
+RUN conda-pack -n assembly -o /tmp/env.tar && \
+    mkdir /venv && cd /venv && tar xf /tmp/env.tar && \
+    rm /tmp/env.tar
+
+RUN /venv/bin/conda-unpack
+
+FROM debian:latest AS runtime
+
+COPY --from=build /venv /venv
+ENV PERL5LIB=/venv/lib/perl5/core_perl
+
+
+ENV PATH="/venv/bin:$PATH"
 RUN git clone --depth 1 https://gitlab.com/chienchi/long_read_assembly.git
 ENV PATH="/long_read_assembly:$PATH"
 
-ADD bin/extractLongReads.pl /opt/conda/bin
-ADD bin/getAvgLen.pl /opt/conda/bin
-ADD bin/renameFilterFasta.pl /opt/conda/bin
-
-CMD ["/bin/bash"]
+SHELL ["/bin/bash", "-c"]
+CMD /bin/bash
diff --git a/runAssembly/bin/renameFilterFasta.pl b/runAssembly/bin/renameFilterFasta.pl
@@ -13,15 +13,17 @@
 my $id_mapping;
 my $contig_size_for_annotation;
 my $project_name;
+my $do_annotation;
 
 GetOptions(
     'u=s{1,}' => \$fasta,
     'd=s' => \$outputDir,
     'filt=i' => \$size_filter,
     'maxseq=i' => \$max_seq_number,
     'id:s' => \$id_mapping,
-    'ann=i' => \$contig_size_for_annotation,
-    'n=s' => \$project_name
+    'ann_size=i' => \$contig_size_for_annotation,
+    'n=s' => \$project_name,
+    'ann:i' => \$do_annotation #default to false (0)
 );
 
 my $output= "$outputDir/${project_name}_contigs.fa";
@@ -56,18 +58,17 @@
         chomp $seq;
         my $fasta_header;
     $id =~ s/\W/_/g;
-        #not doing any of the annotation-specific tasks for this modularized version of runAssembly
-        #if($configuration->{DoAnnotation}){
+        if($do_annotation){
             # genbank format limit the LOCUS name length
-            #if ($id_mapping){
-                #$fasta_header = ( length($id) > 20 ) ? "contig_$serial_id $id" : "$id contig_$serial_id";
-                #print $idmap_ofh "contig_$serial_id\t$id\n";
-            #}else{
-                #$fasta_header = ( length($project_name) > 20 || $project_name =~/^Assembly/i ) ? "contig_$serial_id $id_info GC_content_$GC_content": "${project_name}_$serial_id $id_info GC_content_$GC_content";
-            #}
-        #}else{
+            if ($id_mapping){
+                $fasta_header = ( length($id) > 20 ) ? "contig_$serial_id $id" : "$id contig_$serial_id";
+                print $idmap_ofh "contig_$serial_id\t$id\n";
+            }else{
+                $fasta_header = ( length($project_name) > 20 || $project_name =~/^Assembly/i ) ? "contig_$serial_id $id_info GC_content_$GC_content": "${project_name}_$serial_id $id_info GC_content_$GC_content";
+            }
+        }else{
         $fasta_header = ($id_mapping)? "$id" : "${project_name}_$serial_id $id_info GC_content_$GC_content";
-        #}
+        }
         if ($len >= $contig_size_for_annotation)
         {
         print $ofh2 ">$fasta_header\n" . $seq."\n";

diff --git a/runAssembly/nextflow.config b/runAssembly/nextflow.config
@@ -1,9 +1,14 @@
+process.container = 'apwat/run_assembly:1.4.5'
+singularity  {
+    enabled = true
+    runOptions = "--compat"
+}
 params {
-
     assembler = "IDBA_UD"
     outDir = '.'
     threads = 8 
     projName = "project"
+    annotation = false
     contigSizeForAnnotation = 700
     pairedFiles = "nf_assets/NO_FILE"
     unpairedFile = "nf_assets/NO_FILE2"
@@ -37,6 +42,4 @@ params {
 }
 workflow.onComplete = {
     "rm -rf nf_assets".execute().text
-}
-process.container = 'apwat/run_assembly:1.2.5'
-docker.enabled = true
+}