refactor Mash paste in db build scripts

oschwengers · Jan 11, 2024 · f13fb27 · f13fb27
1 parent acac0c7
commit f13fb27
Show file tree

Hide file tree

Showing 3 changed files with 9 additions and 6 deletions.
diff --git a/db-scripts/build-db-gtdb.sh b/db-scripts/build-db-gtdb.sh
@@ -31,9 +31,10 @@ gunzip metadata.tsv.gz
 DATA_PATH=$(readlink -f gtdb_genomes_reps_r*/database)
 nextflow run $REFERENCE_SEEKER_HOME/db-scripts/build-db-gtdb.nf --metadata ./metadata.tsv --representatives $DATA_PATH --domain $DOMAIN  ||  { echo "Nextflow failed!"; exit; }
 
-mash paste db sketches/*.msh  ||  { echo "Mash failed!"; exit; }
+find sketches/ -type f -name '*.msh' -exec realpath {} + > sketches.fof
+mash paste -l db sketches.fof  ||  { echo "Mash failed!"; exit; }
 
-rm -rf work/ .nextflow* sketches/ *.tsv *_metadata.tar.gz gtdb_genomes_reps.tar.gz
+rm -rf work/ .nextflow* sketches/ sketches.fof metadata.tsv gtdb_genomes_*
 
 mv db.msh $DOMAIN/
 

diff --git a/db-scripts/build-plasmids-db-plsdb.sh b/db-scripts/build-plasmids-db-plsdb.sh
@@ -6,9 +6,10 @@ echo "Unzip plasmid sequences..."
 bunzip2 plsdb.fna.bz2
 nextflow run $REFERENCE_SEEKER_HOME/db-scripts/build-plasmids-db-plsdb.nf --plasmids plsdb.fna  ||  { echo "Nextflow failed!"; exit; }
 
-mash paste db sketches/*.msh  ||  { echo "Mash failed!"; exit; }
+find sketches/ -type f -name '*.msh' -exec realpath {} + > sketches.fof
+mash paste -l db sketches.fof  ||  { echo "Mash failed!"; exit; }
 
-rm -rf work/ .nextflow* sketches/ plsdb.fna
+rm -rf work/ .nextflow* sketches/ sketches.fof plsdb.fna
 
 mv db.msh plasmids-plsdb/
 

diff --git a/db-scripts/build-plasmids-db-refseq.sh b/db-scripts/build-plasmids-db-refseq.sh
@@ -11,9 +11,10 @@ zcat plasmid.* > plasmids.fna
 
 nextflow run $REFERENCE_SEEKER_HOME/db-scripts/build-plasmids-db-refseq.nf --plasmids plasmids.fna  ||  { echo "Nextflow failed!"; exit; }
 
-mash paste db sketches/*.msh  ||  { echo "Mash failed!"; exit; }
+find sketches/ -type f -name '*.msh' -exec realpath {} + > sketches.fof
+mash paste -l db sketches.fof  ||  { echo "Mash failed!"; exit; }
 
-rm -rf work/ .nextflow* sketches/ plasmid.* plasmids.fna
+rm -rf work/ .nextflow* sketches/ plasmid.* sketches.fof plasmids.fna
 
 mv db.msh plasmids-refseq/