Support for tsv files that are either formatted like VCF files or Bed…

… Files (#2) * Support for tsv files that are either formated like VCF files or Bed Files * Add new functional tests
gogetdata · Jul 22, 2020 · 27c154a · 27c154a
1 parent bb63780
commit 27c154a
Show file tree

Hide file tree

Showing 20 changed files with 63 additions and 1 deletion.
diff --git a/.travis.yml b/.travis.yml
@@ -11,6 +11,7 @@ go:
   - 1.12.x
 
 script:
+    - ./functional-test.sh
     - cd cmd/check-sort-order
     - go test
 
diff --git a/cmd/check-sort-order/main.go b/cmd/check-sort-order/main.go
@@ -58,7 +58,9 @@ func main() {
 
 	if strings.HasSuffix(args.Path, ".vcf.gz") {
 		checkVCF(args.Path, gf)
-	} else if strings.HasSuffix(args.Path, ".bed.gz") || strings.HasSuffix(strings.ToLower(args.Path), ".bedgraph.gz") {
+	} else if strings.HasSuffix(args.Path, ".bed.gz") || strings.HasSuffix(strings.ToLower(args.Path), ".bedgraph.gz") || strings.HasSuffix(strings.ToLower(args.Path), ".bedpe.gz"){
+		checkTab(args.Path, gf, get_vcf_chrom_start)
+    } else if strings.HasSuffix(strings.ToLower(args.Path), "tsv.gz"){
 		checkTab(args.Path, gf, get_vcf_chrom_start)
 	} else {
 		found := false

diff --git a/functional-test.sh b/functional-test.sh
@@ -0,0 +1,54 @@
+#!/bin/bash
+
+
+test -e ssshtest || wget -q https://raw.githubusercontent.com/ryanlayer/ssshtest/master/ssshtest
+
+. ssshtest
+
+set -o nounset
+
+go build -o ./check-sort-order cmd/check-sort-order/main.go 
+
+
+## Check vcf file
+run check_vcf_file ./check-sort-order --genome test_genome_files/genome_sizes.genome test_files/test_file.vcf.gz   
+assert_exit_code 0
+assert_no_stdout
+
+## check bed file
+run check_bed_file ./check-sort-order --genome test_genome_files/genome_sizes.genome test_files/test_file.bed.gz   
+assert_exit_code 0
+assert_no_stdout
+
+## check gtf file 
+run check_gtf_file ./check-sort-order --genome test_genome_files/genome_sizes.genome test_files/test_file.gtf.gz   
+assert_exit_code 0
+assert_no_stdout
+
+## Check bad sorted bed 
+run check_bad_sorted_bed_file ./check-sort-order --genome test_genome_files/genome_sizes.genome test_files/test_file.bad_sort.bed.gz  
+assert_exit_code 1
+assert_in_stderr "chromosomes not in specified sort order: 4, 1 at line 33"
+assert_in_stderr "use gsort (https://github.com/brentp/gsort/) to order according to the genome file"
+
+## check a vcf file with a bad header
+run check_bad_sorted_bed_file ./check-sort-order --genome test_genome_files/genome_sizes.genome test_files/test_file.bad_header.vcf.gz
+assert_exit_code 1
+assert_in_stderr "VCF header line '##fileformat=VCF... not found"
+
+## Check a tsv file that is formated like a vcf file
+run check_bad_sorted_bed_file ./check-sort-order --genome test_genome_files/genome_sizes.genome test_files/test_bed_like.tsv.gz
+assert_exit_code 0
+assert_no_stdout
+
+## Check a tsv file that is formated like a bed file
+run check_bad_sorted_bed_file ./check-sort-order --genome test_genome_files/genome_sizes.genome test_files/test_bed_like.tsv.gz
+assert_exit_code 0
+assert_no_stdout
+
+## Check a tsv file that is not formated correctly 
+run check_bad_sorted_bed_file ./check-sort-order --genome test_genome_files/genome_sizes.genome test_files/test_bad.vcf_like.tsv.gz
+assert_exit_code 1
+assert_in_stderr 'strconv.Atoi: parsing "A": invalid syntax'
+
+
diff --git a/test_files/test_bad.vcf_like.tsv.gz b/test_files/test_bad.vcf_like.tsv.gz
diff --git a/test_files/test_bad.vcf_like.tsv.gz.tbi b/test_files/test_bad.vcf_like.tsv.gz.tbi
diff --git a/test_files/test_bed_like.tsv.gz b/test_files/test_bed_like.tsv.gz
diff --git a/test_files/test_bed_like.tsv.gz.tbi b/test_files/test_bed_like.tsv.gz.tbi
diff --git a/test_files/test_file.bad_header.vcf.gz b/test_files/test_file.bad_header.vcf.gz
diff --git a/test_files/test_file.bad_header.vcf.gz.tbi b/test_files/test_file.bad_header.vcf.gz.tbi
diff --git a/test_files/test_file.bad_sort.bed.gz b/test_files/test_file.bad_sort.bed.gz
diff --git a/test_files/test_file.bad_sort.bed.gz.tbi b/test_files/test_file.bad_sort.bed.gz.tbi
diff --git a/test_files/test_file.bed.gz b/test_files/test_file.bed.gz
diff --git a/test_files/test_file.bed.gz.tbi b/test_files/test_file.bed.gz.tbi
diff --git a/test_files/test_file.gtf.gz b/test_files/test_file.gtf.gz
diff --git a/test_files/test_file.gtf.gz.tbi b/test_files/test_file.gtf.gz.tbi
diff --git a/test_files/test_file.vcf.gz b/test_files/test_file.vcf.gz
diff --git a/test_files/test_file.vcf.gz.tbi b/test_files/test_file.vcf.gz.tbi
diff --git a/test_files/test_vcf_like.tsv.gz b/test_files/test_vcf_like.tsv.gz
diff --git a/test_files/test_vcf_like.tsv.gz.tbi b/test_files/test_vcf_like.tsv.gz.tbi
diff --git a/test_genome_files/genome_sizes.genome b/test_genome_files/genome_sizes.genome
@@ -0,0 +1,5 @@
+#chrom	length
+1	249250621
+2	243199373
+3	198022430
+4	191154276