updated testing

malonge · Aug 17, 2020 · db1a34e · db1a34e
1 parent 54e9ba5
commit db1a34e
Show file tree

Hide file tree

Showing 4 changed files with 76 additions and 3 deletions.
diff --git a/tests/integration_tests/.gitignore b/tests/integration_tests/.gitignore
@@ -1,2 +1,3 @@
 ragtag_output_tomato_default
-ragtag_output_Ara_nucmer
+ragtag_output_Ara_nucmer
+ragtag_output_ecoli_val
diff --git a/tests/integration_tests/run_run_tests.sh b/tests/integration_tests/run_run_tests.sh
@@ -1,3 +1,11 @@
 #!/usr/bin/env bash
 
-bash run_tests.sh ~/Projects/Reference_Genomes/Tomato/SL4.0.genome.fasta ~/Projects/ragtag_workspace/tomato_test_env/Brandywine.assembly.polished.fasta ~/Projects/ragtag_workspace/tomato_test_env/brandywine_ctg_genes.gff ~/Projects/Reference_Genomes/Arabidopsis/GCF_000001735.3_TAIR10_genomic.fna ~/Projects/ragtag_workspace/ara_test_env/Arabidopsis.MaSuRCA.3.4.0.fasta ~/Projects/ragtag_workspace/ara_test_env/Arabidopsis.MaSuRCA.3.4.0.genes.gff
+bash run_tests.sh ~/Projects/Reference_Genomes/Tomato/SL4.0.genome.fasta \
+    ~/Projects/ragtag_workspace/tomato_test_env/Brandywine.assembly.polished.fasta \
+    ~/Projects/ragtag_workspace/tomato_test_env/brandywine_ctg_genes.gff \
+    ~/Projects/Reference_Genomes/Arabidopsis/GCF_000001735.3_TAIR10_genomic.fna \
+    ~/Projects/ragtag_workspace/ara_test_env/Arabidopsis.MaSuRCA.3.4.0.fasta \
+    ~/Projects/ragtag_workspace/ara_test_env/Arabidopsis.MaSuRCA.3.4.0.genes.gff \
+    ~/Projects/ragtag_workspace/ecoli_tests/GCF_000008865.2_ASM886v2_genomic.fna \
+    ~/Projects/ragtag_workspace/ecoli_tests/ecoli.contigs.fasta \
+    ~/Projects/ragtag_workspace/ecoli_tests/reads.fofn
diff --git a/tests/integration_tests/run_tests.sh b/tests/integration_tests/run_tests.sh
@@ -7,6 +7,9 @@
 ## 4. ara reference
 ## 5. ara query
 ## 6. ara gff
+## 7. E. coli reference
+## 8. E. coli query
+## 9. E.coli reads fofn
 
 # Assumes fasta suffixes are ".fasta"
 # Assumes gff suffixes are ".gff"
@@ -37,6 +40,11 @@ A_QUERY_PREF=`basename $A_QUERY .fasta`
 A_GFF=$6
 A_GFF_PREF=`basename $A_GFF .gff`
 
+E_REF=$7
+E_QUERY=$8
+E_QUERY_PREF=`basename $E_QUERY .fasta`
+E_VAL_FOFN=$9
+
 
 # Run RagTag
 # Settings: default
@@ -156,4 +164,28 @@ bash scripts/validate_results.sh ragtag_output_Ara_nucmer/ragtag.correction.agp
     ~/Projects/ragtag_workspace/static_results/ara/ragtag.correction.agp
 
 bash scripts/validate_results.sh ragtag_output_Ara_nucmer/ragtag.scaffolds.agp \
-    ~/Projects/ragtag_workspace/static_results/ara/ragtag.scaffolds.agp
+    ~/Projects/ragtag_workspace/static_results/ara/ragtag.scaffolds.agp
+
+
+# Run RagTag
+# Settings: with error-corrected long-read validation
+# Data: E. coli
+echo ""
+echo "**************************************************************"
+echo "***     Running RagTag on E. coli data with validation     ***"
+echo "**************************************************************"
+echo ""
+
+bash scripts/run_val.sh $E_REF \
+    $E_QUERY \
+    $E_VAL_FOFN \
+    ragtag_output_ecoli_val
+
+# Validate the agp files
+echo ""
+mecho "Validating AGP files and associated fasta files:"
+echo ""
+
+bash scripts/validate_agp.sh $E_QUERY \
+    ragtag_output_ecoli_val/$E_QUERY_PREF.corrected.fasta \
+    ragtag_output_ecoli_val/ragtag.correction.agp
diff --git a/tests/integration_tests/scripts/run_val.sh b/tests/integration_tests/scripts/run_val.sh
@@ -0,0 +1,32 @@
+#!/usr/bin/env bash
+
+# run the ragtag correction with read validation (error corrected long reads)
+
+# position args:
+## 1. reference
+## 2. query
+## 3. reads.fofn
+## 4. output dir
+
+# Assumes query suffix is ".fasta"
+
+Usage() {
+    echo "Usage: $0 ref.fa query.fa reads.fofn output_dir"
+}
+
+mecho() {
+    NAME=`basename $0`
+    echo "$NAME:" $1
+}
+
+if [ $# -lt 3 ] ; then
+    Usage
+    exit 1
+fi
+
+REF=$1
+QUERY=$2
+READS=$3
+OUTDIR=$4
+
+ragtag.py correct -F $READS -T corr --debug -t 2 -u -o $OUTDIR $REF $QUERY