Merge pull request #32 from B-UMMI/dev

Add StrainXpress to short-read assembly workflow
B-UMMI · Sep 26, 2022 · fd43031 · fd43031
2 parents 225fba6 + 0929429
commit fd43031
Show file tree

Hide file tree

Showing 12 changed files with 97 additions and 4 deletions.
diff --git a/README.md b/README.md
@@ -173,6 +173,8 @@ To use LMAS the following options are available:
                                     (default auto)
          --skesa                    Boolean controling the execution of the SKESA assembler.
                                     (default true)
+         --strainxpress             Boolean controling the execution of the StrainXpress assembler.
+                                    (default true)
          --unicycler                Boolean controling the execution of the Unicycler assembler.
                                     (default true)
          --velvetoptimiser          Boolean controling the execution of the VelvetOptimiser assembler.

diff --git a/conf/containers.config b/conf/containers.config
@@ -2,6 +2,9 @@ process {
         withName: REFORMAT {
             container = "pcerqueira/bbtools:38.44"
         }
+        withName: REFORMAT_FASTQ {
+            container = "pcerqueira/bbtools:38.44"
+        }
         withName: ABYSS {
             container = "cimendes/abyss:2.3.1-1"
         }
@@ -29,6 +32,9 @@ process {
         withName: SPADES {
             container = "cimendes/spades:3.15.3-1"
         }
+        withName: STRAINXPRESS {
+            container = "cimendes/strainxpress:11.08.2022"
+        }
         withName: UNICYCLER {
             container = "cimendes/unicycler:0.4.9-1"
         }

diff --git a/conf/params.config b/conf/params.config
@@ -46,6 +46,9 @@ params {
         spades = true
         spadesKmerSize = 'auto'
 
+        //StrainXpress
+        strainxpress = false
+
         //SKESA
         skesa = true
 

diff --git a/docker/strainXpress/Dockerfile b/docker/strainXpress/Dockerfile
@@ -0,0 +1,22 @@
+FROM continuumio/miniconda:4.7.12
+
+LABEL software="StrainXpress" \
+    software.version="11.08.2022" \
+    about.home="https://github.com/kangxiongbin/StrainXpress" \
+    about.summary="De novo assembler based on overlap-layout-consensus to assemble high complexity metagenome sequencing data at strain resolution." \
+    about.documentation="https://github.com/kangxiongbin/StrainXpress#readme" \
+    about.license="https://github.com/kangxiongbin/StrainXpress/blob/main/LICENSE" \
+    author="Inês Mendes <[email protected]>"
+
+WORKDIR /NGStools
+
+RUN apt-get --allow-releaseinfo-change update \
+    && apt-get install procps git build-essential cmake wget libboost-all-dev -y 
+
+RUN conda install -c bioconda python=3.6 scipy pandas minimap2
+
+RUN git clone https://github.com/kangxiongbin/StrainXpress.git \
+    && cd StrainXpress \
+    && sh install.sh
+
+ENV PATH="/NGStools/StrainXpress/scripts:$PATH"
diff --git a/lib/Helper.groovy b/lib/Helper.groovy
@@ -103,6 +103,8 @@ class Help {
         println "                               (default $params.spadesKmerSize)"
         println "    --skesa                    Boolean controling the execution of the SKESA assembler."
         println "                               (default $params.skesa)"
+        println "    --strainxpress             Boolean controling the execution of the StrainXpress assembler."
+        println "                               (default $params.strainxpress)"
         println "    --unicycler                Boolean controling the execution of the Unicycler assembler."
         println "                               (default $params.unicycler)"
         println "    --velvetoptimiser          Boolean controling the execution of the VelvetOptimiser assembler."

diff --git a/modules/assembly/assembly.nf b/modules/assembly/assembly.nf
@@ -19,6 +19,24 @@ process REFORMAT {
     "reformat.sh in=${fastq_pair[0]} in2=${fastq_pair[1]} out=${sample_id}_reads.fasta"
 }
 
+process REFORMAT_FASTQ {
+
+    tag { sample_id }
+    label 'process_assembly'
+
+    when:
+    params.strainxpress
+
+    input:
+    tuple val(sample_id), path(fastq_pair)
+
+    output:
+    tuple val(sample_id), file('*_reads.fq') 
+
+    script:
+    "reformat.sh in=${fastq_pair[0]} in2=${fastq_pair[1]} out=${sample_id}_reads.fq"
+}
+
 process ABYSS {
 
     tag { sample_id }
@@ -334,6 +352,38 @@ process SPADES {
     """
 }
 
+process STRAINXPRESS {
+
+    tag { sample_id }
+    label 'process_assembly'
+    publishDir "results/$sample_id/assembly/strainxpress"
+
+    when:
+    params.strainxpress
+
+    input: 
+    tuple val(sample_id), path(fasta_reads_single)
+
+    output:
+    tuple val(sample_id), val('StrainXpress'), path('*_strainxpress.fasta'), emit: assembly
+    path('.*version'), emit: version
+
+    script:
+    """
+    echo '' > .${sample_id}_strainxpress_version
+    {
+        python3 /NGStools/StrainXpress/scripts/strainxpress.py -fq $fasta_reads_single -t $task.cpus
+
+        mv all.contigs_*.fasta ${sample_id}_strainxpress.fasta
+        echo pass > .status
+    } || {
+        echo fail > .status
+        :> ${sample_id}_strainxpress.fasta
+    }
+    """
+
+}
+
 process UNICYCLER {
 
     tag { sample_id }
@@ -419,6 +469,7 @@ workflow assembly_wf {
 
     main:
     REFORMAT(IN_fastq_raw)
+    REFORMAT_FASTQ(IN_fastq_raw)
     ABYSS(IN_fastq_raw, abyssKmerSize, abyssBloomSize)
     GATBMINIAPIPELINE(IN_fastq_raw, gatbKmerSize, GATB_error_correction, gatb_besst_iter)
     IDBA(REFORMAT.out)
@@ -428,6 +479,7 @@ workflow assembly_wf {
     MINIA(IN_fastq_raw, miniaKmerSize)
     SKESA(IN_fastq_raw)
     SPADES(IN_fastq_raw, spadesKmerSize)
+    STRAINXPRESS(REFORMAT_FASTQ.out)
     UNICYCLER(IN_fastq_raw)
     VELVETOPTIMISER(IN_fastq_raw)
 
@@ -440,6 +492,7 @@ workflow assembly_wf {
                                               MINIA.out.assembly,
                                               SKESA.out.assembly,
                                               SPADES.out.assembly,
+                                              STRAINXPRESS.out.assembly,
                                               UNICYCLER.out.assembly,
                                               VELVETOPTIMISER.out.assembly)
     all_versions = ABYSS.out.version | mix(GATBMINIAPIPELINE.out.version,
@@ -450,6 +503,7 @@ workflow assembly_wf {
                                            MINIA.out.version,
                                            SKESA.out.version,
                                            SPADES.out.version,
+                                           STRAINXPRESS.out.version,
                                            UNICYCLER.out.version,
                                            VELVETOPTIMISER.out.version) | collect
 

diff --git a/pull_images.sh b/pull_images.sh
@@ -25,6 +25,7 @@ eval $pull_command"cimendes/mhm2:v2.0.0-65-gaad446d-generic"
 eval $pull_command"cimendes/spades:3.15.3-1"
 eval $pull_command"cimendes/minia:3.2.6-1"
 eval $pull_command"cimendes/skesa:2.5.0-1"
+eval $pull_command"cimendes/strainxpress:11.08.2022"
 eval $pull_command"cimendes/unicycler:0.4.9-1"
 eval $pull_command"cimendes/velvetoptimiser:2.2.6-1"
 eval $pull_command"cimendes/minimap2:2.22-1"
diff --git a/templates/process_assembly_stats_global.py b/templates/process_assembly_stats_global.py
@@ -39,7 +39,8 @@
     ASSEMBLER_SKIP = {"ABySS":json.loads("$params.abyss"), "GATBMiniaPipeline": json.loads("$params.gatb_minia"), 
                        "MetaHipMer2": json.loads("$params.metahipmer2"), "MINIA": json.loads("$params.minia"), "MEGAHIT": json.loads("$params.megahit"), 
                        "metaSPAdes": json.loads("$params.metaspades"), "Unicycler": json.loads("$params.unicycler"), "SPAdes": json.loads("$params.spades"),
-                       "SKESA": json.loads("$params.skesa"), "VelvetOptimiser": json.loads("$params.velvetoptimiser"), "IDBA-UD": json.loads("$params.idba")}
+                       "SKESA": json.loads("$params.skesa"), "StrainXpress": json.loads("$params.strainxpress"),"VelvetOptimiser": json.loads("$params.velvetoptimiser"), 
+                       "IDBA-UD": json.loads("$params.idba")}
     logger.debug("Running {} with parameters:".format(
         os.path.basename(__file__)))
     logger.debug("ASSEMBLY_STATS_GLOBAL_FILE: {}".format(

diff --git a/templates/process_assembly_stats_mapping.py b/templates/process_assembly_stats_mapping.py
@@ -36,7 +36,8 @@
     ASSEMBLER_SKIP = {"ABySS":json.loads("$params.abyss"), "GATBMiniaPipeline": json.loads("$params.gatb_minia"), 
                        "MetaHipMer2": json.loads("$params.metahipmer2"), "MINIA": json.loads("$params.minia"), "MEGAHIT": json.loads("$params.megahit"), 
                        "metaSPAdes": json.loads("$params.metaspades"), "Unicycler": json.loads("$params.unicycler"), "SPAdes": json.loads("$params.spades"),
-                       "SKESA": json.loads("$params.skesa"), "VelvetOptimiser": json.loads("$params.velvetoptimiser"), "IDBA-UD": json.loads("$params.idba")}
+                       "SKESA": json.loads("$params.skesa"), "StrainXpress": json.loads("$params.strainxpress"), "VelvetOptimiser": json.loads("$params.velvetoptimiser"), 
+                       "IDBA-UD": json.loads("$params.idba")}
     logger.debug("Running {} with parameters:".format(
         os.path.basename(__file__)))
     logger.debug("ASSEMBLY_STATS_GLOBAL_FILE_JSON: {}".format(

diff --git a/templates/process_versions.py b/templates/process_versions.py
@@ -45,6 +45,7 @@
     "Unicycler": "UNICYCLER",
     "SPAdes": "SPADES",
     "SKESA": "SKESA",
+    "strainxpress": "StrainXpress",
     "VelvetOptimiser": "VELVETOPTIMISER",
     "IDBA": "IDBA"}
 

diff --git a/templates/utils.py b/templates/utils.py
@@ -15,10 +15,10 @@
            '#fdbf6f', '#ff7f00', '#cab2d6', '#6a3d9a', '#ebdb75', '#b15928']
 
 ASSEMBLER_NAMES = ["ABySS", "GATBMiniaPipeline", "MetaHipMer2", "MINIA", "MEGAHIT", "metaSPAdes", "Unicycler", "SPAdes",
-                   "SKESA", "VelvetOptimiser", "IDBA-UD"]
+                   "StrainXpress", "SKESA", "VelvetOptimiser", "IDBA-UD"]
 
 ASSEMBLER_PROCESS_LIST = ["ABYSS", "GATBMINIAPIPELINE", "MINIA", "METAHIPMER2", "MEGAHIT", "METASPADES", "UNICYCLER", "SPADES",
-                          "SKESA", "VELVETOPTIMISER", "IDBA"]
+                          "STRAINXPRESS", "SKESA", "VELVETOPTIMISER", "IDBA"]
 
 
 def get_logger(filepath, level=logging.DEBUG):

diff --git a/test/data/tiny_reads.fq.gz b/test/data/tiny_reads.fq.gz