feat(metabolomics): carveme independent of cplex

metagenomics · Feb 13, 2024 · 5d7f40c · 5d7f40c
1 parent efd1725
commit 5d7f40c
Show file tree

Hide file tree

Showing 12 changed files with 40 additions and 28 deletions.
diff --git a/.github/workflows/prodTest.yml b/.github/workflows/prodTest.yml
@@ -10,9 +10,8 @@ jobs:
         run: |
           bash ./scripts/test_fullPipeline.sh \
           " -c /vol/spool/aws.config --databases=/vol/scratch/databases/  --logDir s3://toolkit-dev-benchmark \
-           --carveme_image=pbelmann/metabolomics:0.1.0 --steps.metabolomics.beforeProcessScript=/vol/spool/dockerPull.sh \
+           --steps.metabolomics.smetana.beforeProcessScript=/vol/spool/dockerPull.sh \
            --smetana_image=pbelmann/metabolomics:0.1.0 --steps.cooccurrence.beforeProcessScript=/vol/spool/dockerPull.sh \
            --steps.fragmentRecruitment.mashScreen.genomes=/vol/spool/toolkit/gtdb/paths.tsv \
-           --steps.metabolomics.carveme.additionalParams='' " \
           ./default/fullPipeline_illumina_nanpore.yml "" slurm || exit 1
 
diff --git a/.github/workflows/workflow_modules.yml b/.github/workflows/workflow_modules.yml
@@ -48,7 +48,7 @@ jobs:
         run: |
           VERSION=$(sort VERSIONS.txt | tail -n 1)
           bash ./scripts/test_fullPipeline.sh \
-          " --carveme_image=pbelmann/metabolomics:0.1.0  --smetana_image=pbelmann/metabolomics:0.1.0  --databases=/vol/scratch/databases/   --output=output   " \
+          "   --smetana_image=pbelmann/metabolomics:0.1.0  --databases=/vol/scratch/databases/   --output=output   " \
           "" "${WORK_DIR}" ${PROFILE} ${VERSION} || exit 1
 
       - name: Test EMGB import tools
@@ -99,7 +99,7 @@ jobs:
           VERSION=$(sort -r VERSIONS.txt | tail -n 1)
           bash ./scripts/test_fullPipeline.sh \
           " --databases=/vol/scratch/databases/ --output=output \
-           --smetana_image=pbelmann/metabolomics:0.1.0 --carveme_image=pbelmann/metabolomics:0.1.0 --steps.cooccurrence.beforeProcessScript=/vol/spool/dockerPull.sh  " \
+           --smetana_image=pbelmann/metabolomics:0.1.0 --steps.cooccurrence.beforeProcessScript=/vol/spool/dockerPull.sh  " \
           ./example_params/fullPipelineIlluminaOrONT.yml \
           "${WORK_DIR}" ${PROFILE} ${VERSION} || exit 1
 
@@ -377,8 +377,9 @@ jobs:
       - name: Metabolomics
         run: |
           bash ./scripts/test_metabolomics.sh "   -c  test_data/assets/aws.config     \
-                                               --carveme_image=pbelmann/metabolomics:0.1.0 --steps.metabolomics.beforeProcessScript=/vol/spool/dockerPull.sh \
-                                               --smetana_image=pbelmann/metabolomics:0.1.0 --gapseq_image=pbelmann/metabolomics:0.1.0 --steps.metabolomics.carveme.additionalParams=''  " \
+                                               --steps.metabolomics.smetana.beforeProcessScript=/vol/spool/dockerPull.sh \
+                                               --steps.metabolomics.memote.beforeProcessScript=/vol/spool/dockerPull.sh \
+                                               --smetana_image=pbelmann/metabolomics:0.1.0   " \
                                                "" "${WORK_DIR}" "slurm"
   plasmid:
     timeout-minutes: 800

diff --git a/default/fullPipeline_illumina_nanpore.yml b/default/fullPipeline_illumina_nanpore.yml
@@ -289,19 +289,20 @@ steps:
           source: s3://databases_internal/kegg-links-mirror-2021-01.tar.gz
           md5sum: 69538151ecf4128f4cb1e0ac21ff55b3
   metabolomics:
-     beforeProcessScript: ""
      filter:
        minCompleteness: 50
        maxContamination: 5
-     gapseq:
-       additionalParams: ""
+     carveme:
+       additionalParams: " --solver scip "
      smetana:
+       beforeProcessScript: ""
        global: false
        detailed: false
        additionalParams:
          detailed: ""
          global: ""
      memote:
+       beforeProcessScript: ""
        additionalParams:
          run: ""
          report: ""

diff --git a/default/fullPipeline_illumina_nanpore_without_aggregate.yml b/default/fullPipeline_illumina_nanpore_without_aggregate.yml
@@ -248,19 +248,20 @@ steps:
           source: https://openstack.cebitec.uni-bielefeld.de:8080/databases/mob_20220929.gz
           md5sum: 21fcaf9c3754a985d1d6875939d71e28
   metabolomics:
-     beforeProcessScript: ""
      filter:
        minCompleteness: 50
        maxContamination: 5
-     gapseq:
-       additionalParams: ""
+     carveme:
+       additionalParams: " --solver scip "
      smetana:
+       beforeProcessScript: ""
        global: false
        detailed: false
        additionalParams:
          detailed: ""
          global: ""
      memote:
+       beforeProcessScript: ""
        additionalParams:
          run: ""
          report: ""

diff --git a/docker/toolkit-carveme/Dockerfile b/docker/toolkit-carveme/Dockerfile
@@ -0,0 +1,7 @@
+FROM quay.io/metagenomics/toolkit-python:3.7-1
+RUN pip install carveme==1.6.1 pyscipopt \
+	&& wget http://github.com/bbuchfink/diamond/releases/download/v2.1.9/diamond-linux64.tar.gz \
+	&& tar xzvf diamond-linux64.tar.gz \
+	&& chmod a+x diamond \
+	&& mv diamond /usr/local/bin/
+
diff --git a/docker/toolkit-carveme/VERSION b/docker/toolkit-carveme/VERSION
@@ -0,0 +1 @@
+1.6.1-0
diff --git a/docs/modules/metabolomics.md b/docs/modules/metabolomics.md
@@ -33,7 +33,7 @@ In the following example your the image name is metabolomics:0.1.0:
 
 * `--gapseq_image=metabolomics:0.1.0` (Optional)
 * `--smetana_image=metabolomics:0.1.0` (Required)
-* `--carveme_image=metabolomics:0.1.0` (Required)
+* `--carveme_image=metabolomics:0.1.0` (Optional. Carveme is not able to detect the solver automatically. Please specify `--solver` in the configuration file if you want to use the scip solver.)
 * `--memote_image=metabolomics:0.1.0` (Optional. Memote is not able to detect the solver automatically. Please specify `--solver` in the configuration file if you are not using the glpk solver.)
 
 For gapseq and memote we are using a publicly available docker image that uses the freely available glkp solver which means that you don't have to provide this parameter.

diff --git a/example_params/fullPipeline.yml b/example_params/fullPipeline.yml
@@ -217,19 +217,20 @@ steps:
       database:
         extractedDBPath: "/vol/spool/toolkit/kegg_2021-01/" 
   metabolomics:
-     beforeProcessScript: /vol/spool/dockerPull.sh
      filter:
        minCompleteness: 49
        maxContamination: 5
      carveme:
-       additionalParams: ""
+       additionalParams: " --solver scip "
      smetana:
+       beforeProcessScript: /vol/spool/dockerPull.sh
        global: false
        detailed: false
        additionalParams:
          detailed: ""
          global: ""
      memote:
+       beforeProcessScript: /vol/spool/dockerPull.sh
        additionalParams:
          run: ""
          report: ""

diff --git a/example_params/metabolomics.yml b/example_params/metabolomics.yml
@@ -13,21 +13,22 @@ logLevel: 1
 scratch: "/vol/scratch"
 steps:
   metabolomics:
-     beforeProcessScript: /vol/spool/dockerPull.sh
      input:
        bins: "test_data/metabolomics/input.tsv" 
      filter:
        minCompleteness: 49
        maxContamination: 5
      carveme:
-       additionalParams: ""
+       additionalParams: " --solver scip "
      smetana:
+       beforeProcessScript: /vol/spool/dockerPull.sh
        global: true
        detailed: true
        additionalParams:
          detailed: ""
          global: ""
      memote:
+       beforeProcessScript: /vol/spool/dockerPull.sh
        additionalParams:
          run: ""
          report: ""

diff --git a/modules/metabolomics/module.nf b/modules/metabolomics/module.nf
@@ -17,8 +17,8 @@ process pGapSeq {
 
     tag "Sample: $sample, Bin: $id"
 
-    beforeScript params?.steps.containsKey("metabolomics") \
-	? Utils.getBeforeScript(params?.steps?.metabolomics?.beforeProcessScript.trim(), params.gapseq_image) \
+    beforeScript params?.steps.containsKey("metabolomics")  && params?.steps?.metabolomics?.gapseq?.containsKey("beforeProcessScript") \
+	? Utils.getBeforeScript(params?.steps?.metabolomics?.gapseq?.beforeProcessScript.trim(), params.gapseq_image) \
 	: ""
 
     when params.steps.containsKey("metabolomics") \
@@ -60,8 +60,8 @@ process pMemote {
 
     container "${params.memote_image}"
 
-    beforeScript params?.steps.containsKey("metabolomics") \
-	? Utils.getBeforeScript(params?.steps?.metabolomics?.beforeProcessScript.trim(), params.memote_image) \
+    beforeScript params?.steps.containsKey("metabolomics") && params?.steps?.metabolomics?.memote?.containsKey("beforeProcessScript") \
+	? Utils.getBeforeScript(params?.steps?.metabolomics?.memote?.beforeProcessScript.trim(), params.memote_image) \
 	: ""
 
     when params.steps.containsKey("metabolomics") && params.steps.metabolomics.containsKey("memote")
@@ -92,8 +92,8 @@ process pSmetanaDetailed {
 
     container "${params.smetana_image}"
 
-    beforeScript params?.steps.containsKey("metabolomics") \
-	? Utils.getBeforeScript(params?.steps?.metabolomics?.beforeProcessScript.trim(), params.smetana_image) \
+    beforeScript params?.steps.containsKey("metabolomics") && params?.steps?.metabolomics?.smetana?.containsKey("beforeProcessScript")  \
+	? Utils.getBeforeScript(params?.steps?.metabolomics?.smetana?.beforeProcessScript.trim(), params.smetana_image) \
 	: ""
 
     publishDir params.output, mode: "${params.publishDirMode}", saveAs: { filename -> getOutput("${sample}", params.runid , "smetana/detailed/", filename) }, \
@@ -121,8 +121,8 @@ process pSmetanaGlobal {
 
     tag "Sample: $sample"
 
-    beforeScript params?.steps.containsKey("metabolomics") \
-	? Utils.getBeforeScript(params?.steps?.metabolomics?.beforeProcessScript.trim(), params.smetana_image) \
+    beforeScript params?.steps.containsKey("metabolomics") && params?.steps?.metabolomics?.smetana?.containsKey("beforeProcessScript") \
+	? Utils.getBeforeScript(params?.steps?.metabolomics?.smetana?.beforeProcessScript.trim(), params.smetana_image) \
 	: ""
 
     container "${params.smetana_image}"

diff --git a/modules/metabolomics/processes.nf b/modules/metabolomics/processes.nf
@@ -18,8 +18,8 @@ process pCarveMe {
 
     when params.steps.containsKey("metabolomics") && params.steps.metabolomics.containsKey("carveme")
 
-    beforeScript  params?.steps.containsKey("metabolomics") \
-	? Utils.getBeforeScript(params?.steps?.metabolomics?.beforeProcessScript.trim(), params.carveme_image) \
+    beforeScript  params?.steps.containsKey("metabolomics") && params?.steps?.metabolomics?.carveme?.containsKey("beforeProcessScript")  \
+	? Utils.getBeforeScript(params.steps.metabolomics.carveme.beforeProcessScript.trim(), params.carveme_image) \
 	: ""
 
     publishDir params.output, mode: "${params.publishDirMode}", saveAs: { filename -> getOutput("${sample}", params.runid, "carveme", filename) }, \

diff --git a/nextflow.config b/nextflow.config
@@ -266,7 +266,7 @@ params {
    bandage_image = "quay.io/biocontainers/bandage:0.8.1--hc9558a2_2"
    nonpareil_image = "quay.io/biocontainers/nonpareil:3.4.1--r41h9f5acd7_1"
    ubuntu_image = "quay.io/metagenomics/toolkit-ubuntu:20.04-0"
-   carveme_image = "carveme"
+   carveme_image = "quay.io/metagenomics/toolkit-carveme:1.6.1-0"
    cobra_image = "quay.io/biocontainers/cobra:0.21.0--pyh5e36f6f_1"
    memote_image = "quay.io/metagenomics/toolkit-memote:0.13.0-0"
    smetana_image = "smetana"