fix merge conflicts

metagenomics · Oct 16, 2023 · 6165fd9 · 6165fd9
2 parents beaa08e + dc1cb42
commit 6165fd9
Show file tree

Hide file tree

Showing 4 changed files with 17 additions and 21 deletions.
diff --git a/.github/workflows/workflow_modules.yml b/.github/workflows/workflow_modules.yml
@@ -43,9 +43,14 @@ jobs:
         run: |
           VERSION=$(sort VERSIONS.txt | tail -n 1)
           bash ./scripts/test_fullPipeline.sh \
-          " --databases=/vol/scratch/databases/      " \
+          " --databases=/vol/scratch/databases/   --output=output   " \
           "" "${WORK_DIR}" ${PROFILE} ${VERSION} || exit 1
 
+      - name: Test EMGB import tools
+        run: |
+          ./bin/emgb.sh --output=output/test1 --runid=1 --binsdir=$(find output/test1/ -name "metabat") \
+                        --workdir="${WORK_DIR}_wFullPipeline" --blastdb=bacmet20_predicted --name=test1
+
       - name: Test Full Pipeline run that ends with magAttributes
         run: |
           bash ./scripts/test_fullPipeline.sh \
@@ -99,8 +104,6 @@ jobs:
           bash ./scripts/test_fullPipeline.sh   " -c  test_data/assets/aws.config     "  \
           ./example_params/fullPipelineSRAQCONT.yml "${WORK_DIR}" ${PROFILE} || exit 1
 
-
-
   mag_attributes:
     timeout-minutes: 800
     needs: [codePolicy]

diff --git a/bin/emgb.sh b/bin/emgb.sh
@@ -16,6 +16,8 @@ while [ $# -gt 0 ]; do
 	    ;;
 	    --blastdb=*) BLAST_DB="${1#*=}"
 	    ;;
+	    --workdir=*) WORK_DIR="${1#*=}"
+	    ;;
 	    --version) VERSION_CHECK=1
 	    ;;
 	    --debug) DEBUG_CHECK=1
@@ -52,7 +54,7 @@ function getGenes {
 		echo $cmd
 	fi
 
-	docker run -i -v $(pwd):$(pwd) -v ${OUTPUT_PATH}:${OUTPUT_PATH} quay.io/emgb/annotatedgenes2json:2.2.2 $cmd
+	docker run -i -v $(pwd):$(pwd) -v $WORK_DIR:$WORK_DIR -v ${OUTPUT_PATH}:${OUTPUT_PATH} quay.io/emgb/annotatedgenes2json:2.2.2 $cmd
 }
 
 
@@ -71,7 +73,7 @@ function getContigs {
 		echo $cmd
 	fi
 
-	docker run -i -v $(pwd):$(pwd) -v ${OUTPUT_PATH}:${OUTPUT_PATH} quay.io/emgb/annotatedcontigs2json:2.2.2 $cmd
+	docker run -i -v $(pwd):$(pwd) -v $WORK_DIR:$WORK_DIR -v ${OUTPUT_PATH}:${OUTPUT_PATH} quay.io/emgb/annotatedcontigs2json:2.2.2 $cmd
 }
 
 
@@ -81,14 +83,14 @@ function getBins {
 	bins=$(find $BINS_DIR -name "*_bin.*.fa" -exec readlink -f {} \; | tail -n 1 | rev | cut -f 1 -d '/' | rev | cut -d '.' -f 1 | sed 's/^/  -bin-id-prefix /g')
 	json=" -json-gz $(pwd)/${NAME}.bins.json.gz "
 
-	$cmd="$checkm $gtdbtk $bins. $json"
+	cmd="$checkm $gtdbtk $bins. $json"
 
 	if [ ! -z "$DEBUG_CHECK" ]
 	then
 		echo $cmd
 	fi
 
-	docker run -i -v $(pwd):$(pwd) -v ${OUTPUT_PATH}:${OUTPUT_PATH} quay.io/emgb/annotatedbins2json:2.2.2 $cmd
+	docker run -i -v $(pwd):$(pwd) -v $WORK_DIR:$WORK_DIR -v ${OUTPUT_PATH}:${OUTPUT_PATH} quay.io/emgb/annotatedbins2json:2.2.2 $cmd
 }
 
 
@@ -104,6 +106,7 @@ help()
 	echo "              -- (e.g. the folder name of BLAST_DB: output/test1/1/annotation/0.3.0/mmseqs2/BLAST_DB)"
 	echo "              -- (Examples: bacmet20_predicted, ncbi_nr)"
 	echo "  --db        -- emgb specific kegg database"
+	echo "  --workdir   -- absolute path to Nextflow work directory"
 	echo "  --help      -- help page"
 	echo "  --debug     -- print commands before running"
 	echo "  --version   -- version of this script"

diff --git a/example_params/fullPipeline.yml b/example_params/fullPipeline.yml
@@ -200,14 +200,6 @@ steps:
           md5sum: d7e627221a1d4584d1c1795cda774cdb
     mmseqs2_taxonomy:
       runOnMAGs: true
-      gtdb:
-        params: ' --orf-filter-s 1 -e 1e-15'
-        ramMode: false
-        initialMaxSensitivity: 1
-        database:
-          download:
-            source: https://openstack.cebitec.uni-bielefeld.de:8080/databases/gtdb_r214_1_mmseqs.tar.gz
-            md5sum: 3c8f12c5c2dc55841a14dd30a0a4c718
       ncbi_nr:
         params: ' --orf-filter-s 1 -e 1e-15'
         ramMode: false
@@ -220,13 +212,11 @@ steps:
       database:
         extractedDBPath: "/vol/spool/toolkit/kegg_2021-01/" 
   metabolomics:
-     beforeProcessScript: ""
-     input:
-       bins: "test_data/metabolomics/input.tsv" 
+     beforeProcessScript: /vol/spool/dockerPull.sh
      filter:
        minCompleteness: 49
        maxContamination: 5
-     gapseq:
+     carveme:
        additionalParams: ""
      smetana:
        global: false

diff --git a/example_params/metabolomics.yml b/example_params/metabolomics.yml
@@ -12,13 +12,13 @@ logLevel: 1
 scratch: "/vol/scratch"
 steps:
   metabolomics:
-     beforeProcessScript: ""
+     beforeProcessScript: /vol/spool/dockerPull.sh
      input:
        bins: "test_data/metabolomics/input.tsv" 
      filter:
        minCompleteness: 49
        maxContamination: 5
-     gapseq:
+     carveme:
        additionalParams: ""
      smetana:
        global: true