[WIP] LONGLMAS: First working iteration PR

.github/workflows/ci_nextflow.yml

@@ -28,6 +28,15 @@ jobs:
           --velvetoptimiser false
           FILE=report/index.html 
           if [ ! -f "$FILE" ]; then echo "Run failed!"; exit 1; else echo "Run successful!"; fi
+          rm -r work/ results/ report/ .nextflow.log pipeline_stats.txt
+      - name: Basic workflow tests - long 
+        run: |
+          nextflow run ${GITHUB_WORKSPACE} -profile docker --fastq "test/data/tinyont.fastq.gz" \
+          --reference "test/data/tiny_reference.fasta" --max_cpus 2 --max_memory 4.GB \
+          --wf ont --flye false --metaflye false --ra false --wtdbg2 false --canu false
+          FILE=report/index.html 
+          if [ ! -f "$FILE" ]; then echo "Run failed!"; exit 1; else echo "Run successful!"; fi
+          rm -r work/ results/ report/ .nextflow.log pipeline_stats.txt
 
   singularity:
     runs-on: ubuntu-latest
@@ -54,4 +63,13 @@ jobs:
           --minia false --megahit false --metaspades false --spades false --unicycler false \
           --velvetoptimiser false
           FILE=report/index.html 
-          if [ ! -f "$FILE" ]; then echo "Run failed!"; exit 1; else echo "Run successful!"; fi
+          if [ ! -f "$FILE" ]; then echo "Run failed!"; exit 1; else echo "Run successful!"; fi
+          rm -r work/ results/ report/ .nextflow.log pipeline_stats.txt
+      - name: Basic workflow tests - long 
+        run: |
+          nextflow run ${GITHUB_WORKSPACE} -profile singularity --fastq "test/data/tinyont.fastq.gz" \
+          --reference "test/data/tiny_reference.fasta" --max_cpus 2 --max_memory 4.GB \
+          --wf ont --flye false --metaflye false --ra false --wtdbg2 false --canu false
+          FILE=report/index.html 
+          if [ ! -f "$FILE" ]; then echo "Run failed!"; exit 1; else echo "Run successful!"; fi
+          rm -r work/ results/ report/ .nextflow.log pipeline_stats.txt

conf/containers.config

@@ -8,6 +8,12 @@ process {
         withName: ABYSS {
             container = "cimendes/abyss:2.3.1-1"
         }
+        withName: CANU {
+            container = "cimendes/canu:2.2"
+        }
+        withName: FLYE {
+            container = "cimendes/flye:2.9"
+        }
         withName: GATBMINIAPIPELINE {
             container = "cimendes/gatb-minia-pipeline:31.07.2020-1"
         }
@@ -17,6 +23,9 @@ process {
         withName: MEGAHIT {
             container = "cimendes/megahit-assembler:1.2.9-1"
         }
+        withName: METAFLYE {
+            container = "cimendes/flye:2.9"
+        }
         withName: METAHIPMER2 {
             container = "cimendes/mhm2:v2.0.0-65-gaad446d-generic"
         }
@@ -26,6 +35,12 @@ process {
         withName: MINIA {
             container = "cimendes/minia:3.2.6-1"
         }
+        withName: RA {
+            container = "cimendes/ra:0.2.1"
+        }
+        withName: RAVEN {
+            container = "cimendes/raven:1.8.0"
+        }
         withName: SKESA {
             container = "cimendes/skesa:2.5.0-1"
         }
@@ -41,6 +56,9 @@ process {
         withName: VELVETOPTIMISER {
             container = "cimendes/velvetoptimiser:2.2.6-1"
         }
+        withName: WTDBG2 {
+            container = "cimendes/wtdgb2:2.5"
+        }
         withName: FILTER_ASSEMBLY {
             container = "pcerqueira/bbtools:38.44"
         }
@@ -53,4 +71,7 @@ process {
         withName: READ_MAPPING {
             container = "cimendes/minimap2:2.22-1"
         }
+        withName: LONG_READ_MAPPING {
+            container = "cimendes/minimap2:2.22-1"
+        }
 }

conf/params.config

@@ -1,10 +1,17 @@
 params {
 
+        /*
+        Run Mode
+        ---------------------------
+        */
+        wf = "default"
+
         /*
         Input Files
         ---------------------------
         */
         fastq = 'data/fastq/*_{1,2}.*'
+        ont = false
         reference = 'data/reference/*.fasta'
         md = 'data/*.md'
 
@@ -60,6 +67,29 @@ params {
         velvetoptimiser_hashs = 19
         velvetoptimiser_hashe = 31
 
+        /*
+        Long-Read Assemblers
+        ---------------------------
+        */
+        //CANU
+        canu = false
+        canu_genomesize = "1g"
+
+        //FLYE
+        flye = true
+
+        //METAFLYE
+        metaflye = true
+
+        //RA
+        ra = true
+
+        //RAVEN
+        raven = true
+
+        //WTDBG2
+        wtdbg2 = true
+
         /*
         Assembly Quality Control
         */

docker/canu/Dockerfile

@@ -0,0 +1,18 @@
+FROM continuumio/miniconda3
+
+LABEL software="CANU" \
+    software.version="2.2" \
+    about.home="https://github.com/marbl/canu" \
+    about.summary="A single molecule sequence assembler for genomes large and small." \
+    about.documentation="https://canu.readthedocs.io/en/latest/index.html" \
+    about.license="https://github.com/marbl/canu/blob/master/README.licenses" \
+    author="Inês Mendes <[email protected]>"
+
+# System essentials
+RUN conda install -c conda-forge -c bioconda -c defaults canu=2.2
+
+# Set up workdir for installation 
+WORKDIR /data
+
+# PS for nextflow
+RUN apt-get install -y procps

docker/flye/Dockerfile

@@ -0,0 +1,16 @@
+FROM continuumio/miniconda3
+
+LABEL software="Flye" \
+    software.version="2.9" \
+    about.home="https://github.com/fenderglass/Flye" \
+    about.summary="Flye Assembler for long read data" \
+    author="Inês Mendes <[email protected]>"
+
+# System essentials
+RUN conda install -c conda-forge -c bioconda -c defaults flye=2.9
+
+# Set up workdir for installation 
+WORKDIR /data
+
+# PS for nextflow
+RUN apt-get install -y procps

docker/miniasm/Dockerfile

@@ -0,0 +1,14 @@
+# Base image - Alternatives: Conda, Python...
+FROM continuumio/miniconda3
+
+LABEL software="miniasm" \
+    software.version="0.3_r179" \
+    about.home="https://github.com/lh3/miniasm" \
+    about.summary="Ultrafast de novo assembly for long noisy reads (though having no consensus step)" \
+    author="Inês Mendes <[email protected]>"
+
+# System essentials
+RUN conda install -c bioconda miniasm
+
+# PS for nextflow
+RUN apt-get install -y procps

docker/ra/Dockerfile

@@ -0,0 +1,26 @@
+# Base image - Alternatives: Conda, Python...
+FROM continuumio/miniconda3
+
+LABEL software="ra" \
+    software.version="0.2.1" \
+    about.home="https://github.com/lbcb-sci/ra" \
+    about.summary="Assembler for raw de novo genome assembly of long uncorrected reads." \
+    about.license="https://github.com/lbcb-sci/ra/blob/master/LICENSE" \
+    author="Inês Mendes <[email protected]>"
+
+# System essentials
+RUN conda install -c bioconda ra
+RUN apt-get update && apt-get -y install build-essential \
+                                         cmake \
+                                         zlib1g-dev
+
+# Install RA
+RUN git clone --recursive https://github.com/rvaser/ra.git ra
+WORKDIR ra
+RUN mkdir build && cd build && cmake -DCMAKE_BUILD_TYPE=Release ..
+RUN cd build && make
+
+ENV PATH /ra/build/bin:$PATH
+
+# PS for nextflow
+RUN apt-get install -y procps

docker/raven/Dockerfile

@@ -0,0 +1,14 @@
+# Base image - Alternatives: Conda, Python...
+FROM continuumio/miniconda3
+
+LABEL software="raven-assembler" \
+    software.version="1.8.0" \
+    about.home="https://github.com/lbcb-sci/raven" \
+    about.summary="Raven-assembler for long read data" \
+    author="Inês Mendes <[email protected]>"
+
+# System essentials
+RUN conda install -c conda-forge -c bioconda -c defaults raven-assembler=1.8.0
+
+# PS for nextflow
+RUN apt-get install -y procps

docker/shaste/Dockerfile

@@ -0,0 +1,14 @@
+# Base image - Alternatives: Conda, Python...
+FROM continuumio/miniconda3
+
+LABEL software="shasta" \
+    software.version="0.6.0" \
+    about.home="https://github.com/chanzuckerberg/shasta" \
+    about.summary="De novo assembly from Oxford Nanopore reads." \
+    author="Inês Mendes <[email protected]>"
+
+# System essentials
+RUN conda install -c bioconda shasta
+
+# PS for nextflow
+RUN apt-get install -y procps

docker/wtdgb2/Dockerfile

@@ -0,0 +1,17 @@
+# Base image - Alternatives: Conda, Python...
+FROM continuumio/miniconda3
+
+LABEL software="wtdbg2" \
+    software.version="2.5" \
+    about.home="https://github.com/ruanjue/wtdbg2" \
+    about.summary="wtdbg2 assembler for long read data" \
+    author="Inês Mendes <[email protected]>"
+
+# Maintainer
+LABEL maintainer="Inês Mendes <[email protected]>"
+
+# System essentials
+RUN conda install -c conda-forge -c bioconda -c defaults wtdbg=2.5
+
+# PS for nextflow
+RUN apt-get install -y procps

lib/CheckParams.groovy

@@ -1,6 +1,14 @@
 class Params {
 
     static void check(Map params) {
+
+        // mode
+        if (!params.wf) {print_error("'--wf' parameter missing")}
+        def allowed_modes = ["default", "Illumina", "illumina", "hybrid", "Hybrid", "ont", "ONT"] as Set
+        def mode_parameter_diff = allowed_modes - params.wf
+        if (mode_parameter_diff.size() > 6){
+                print_error("[Pipeline warning] Parameter --wf $params.wf is not valid in the pipeline! Allowed values: 'default', 'long', 'hybrid', 'all'")
+    }
 
         // input
         if (!params.reference) {print_error("'--reference' parameter missing")}

lib/Helper.groovy

@@ -33,8 +33,12 @@ class Help {
         println "         Last Metagenomic Assembler Standing"
         println ""
         println "Input parameters:"
-        println "    --fastq                    Path expression to paired-end fastq files." 
+        println "    --wf                       Execution mode of LMAS."
+        println "                               (default: $params.wf, allowed options: 'default' or 'illumina', 'ont')"
+        println "    --fastq                    Path expression to short-read paired-end fastq files." 
         println "                               (default: $params.fastq)"
+        println "    --ont                      Path expression to nanopore fastq files." 
+        println "                               (default: $params.ont)"
         println "    --reference                Path to the genome reference fasta file."
         println "                               (default: $params.reference)"
         println "    --md                       Path to markdown with input sample description for report (optional)."
@@ -56,7 +60,7 @@ class Help {
         println "                               Allowed values: 'linear' or 'log'."
         println "                               (default: $params.plot_scale)"
         println ""
-        println "Assembly execution parameters:"
+        println "Short-read assembly execution parameters:"
         println "    --abyss                    Boolean controling the execution of the ABySS assembler."
         println "                               (default: $params.abyss)"
         println "    --abyssKmerSize            K-mer size for the ABySS assembler, as an intiger."
@@ -114,6 +118,22 @@ class Help {
         println "    --velvetoptimiser_hashe    End K-mer size for the VelvetOptimiser assembler, as an intiger."
         println "                               (default $params.velvetoptimiser_hashe)"
         println ""
+        println "Long-read assembly execution parameters:"
+        println "    --canu                     Boolean controling the execution of the canu assembler."
+        println "                               (default: $params.canu)"
+        println "    --canu_genomesize          An estimate of the size of the genome for the canu assembler."
+        println "                               (default: $params.canu_genomesize)"
+        println "    --flye                     Boolean controling the execution of the flye assembler."
+        println "                               (default: $params.flye)"
+        println "    --metaflye                 Boolean controling the execution of the metaflye assembler."
+        println "                               (default: $params.metaflye)"
+        println "    --ra                       Boolean controling the execution of the ra assembler."
+        println "                               (default: $params.ra)"
+        println "    --raven                    Boolean controling the execution of the raven assembler."
+        println "                               (default: $params.raven)"
+        println "    --wtdbg2                   Boolean controling the execution of the wtdbg2 assembler."
+        println "                               (default: $params.wtdbg2)"
+        println ""
         println "Execution resources parameters:"
         println "    --cpus                     Number of CPUs for the assembly and mapping processes, as an intiger."
         println "                               This resource is double for each retry until max_cpus is reached."