Reorganise the test data, add readgroups test

tests/data/config/invalid_config_fake_ref.json

@@ -3,13 +3,13 @@
     "micro": {
       "libraries": {
         "lib_01": {
-        "R1": "tests/data/micro_R1.fq.gz",
-        "R2": "tests/data/micro_R2.fq.gz"
+        "R1": "tests/data/fastq/micro_R1.fq.gz",
+        "R2": "tests/data/fastq/micro_R2.fq.gz"
         }
       }
     }
   },
-  "dbsnp": "tests/data/database.vcf.gz",
-  "known_sites": ["tests/data/database.vcf.gz"],
+  "dbsnp": "tests/data/reference/database.vcf.gz",
+  "known_sites": ["tests/data/reference/database.vcf.gz"],
   "reference": "/this/file/is/not/there"
 }

tests/data/config/invalid_config_no_ref.json

@@ -3,12 +3,12 @@
     "micro": {
       "libraries": {
         "lib_01": {
-        "R1": "tests/data/micro_R1.fq.gz",
-        "R2": "tests/data/micro_R2.fq.gz"
+        "R1": "tests/data/fastq/micro_R1.fq.gz",
+        "R2": "tests/data/fastq/micro_R2.fq.gz"
         }
       }
     }
   },
-  "dbsnp": "tests/data/database.vcf.gz",
-  "known_sites": ["tests/data/database.vcf.gz"]
+  "dbsnp": "tests/data/reference/database.vcf.gz",
+  "known_sites": ["tests/data/reference/database.vcf.gz"]
 }

tests/data/config/sample_config.json

@@ -3,13 +3,13 @@
     "micro": {
       "libraries": {
         "lib_01": {
-        "R1": "tests/data/micro_R1.fq.gz",
-        "R2": "tests/data/micro_R2.fq.gz"
+        "R1": "tests/data/fastq/micro_R1.fq.gz",
+        "R2": "tests/data/fastq/micro_R2.fq.gz"
         }
       }
     }
   },
-  "reference":"tests/data/ref.fa",
-  "dbsnp": "tests/data/database.vcf.gz",
-  "known_sites": ["tests/data/database.vcf.gz"]
+  "reference":"tests/data/reference/ref.fa",
+  "dbsnp": "tests/data/reference/database.vcf.gz",
+  "known_sites": ["tests/data/reference/database.vcf.gz"]
 }

tests/data/config/sample_config_bed_all.json

@@ -3,14 +3,14 @@
     "micro": {
       "libraries": {
         "lib_01": {
-        "R1": "tests/data/micro_R1.fq.gz",
-        "R2": "tests/data/micro_R2.fq.gz"
+        "R1": "tests/data/fastq/micro_R1.fq.gz",
+        "R2": "tests/data/fastq/micro_R2.fq.gz"
         }
       }
     }
   },
-  "reference":"tests/data/ref.fa",
-  "dbsnp": "tests/data/database.vcf.gz",
-  "known_sites": ["tests/data/database.vcf.gz"],
-  "bedfile": "tests/data/full_chrM.bed"
+  "reference":"tests/data/reference/ref.fa",
+  "dbsnp": "tests/data/reference/database.vcf.gz",
+  "known_sites": ["tests/data/reference/database.vcf.gz"],
+  "bedfile": "tests/data/reference/full_chrM.bed"
 }

tests/data/config/sample_config_refflat.json

@@ -3,14 +3,14 @@
     "micro": {
       "libraries": {
         "lib_01": {
-        "R1": "tests/data/micro_R1.fq.gz",
-        "R2": "tests/data/micro_R2.fq.gz"
+        "R1": "tests/data/fastq/micro_R1.fq.gz",
+        "R2": "tests/data/fastq/micro_R2.fq.gz"
         }
       }
     }
   },
-  "reference":"tests/data/ref.fa",
-  "dbsnp": "tests/data/database.vcf.gz",
-  "known_sites": ["tests/data/database.vcf.gz"],
-  "refflat": "tests/data/chrM_refFlat.txt"
+  "reference":"tests/data/reference/ref.fa",
+  "dbsnp": "tests/data/reference/database.vcf.gz",
+  "known_sites": ["tests/data/reference/database.vcf.gz"],
+  "refflat": "tests/data/reference/chrM_refFlat.txt"
 }

tests/data/config/sample_config_scatter.json

@@ -3,14 +3,14 @@
     "micro": {
       "libraries": {
         "lib_01": {
-        "R1": "tests/data/micro_R1.fq.gz",
-        "R2": "tests/data/micro_R2.fq.gz"
+        "R1": "tests/data/fastq/micro_R1.fq.gz",
+        "R2": "tests/data/fastq/micro_R2.fq.gz"
         }
       }
     }
   },
-  "reference":"tests/data/ref.fa",
-  "dbsnp": "tests/data/database.vcf.gz",
-  "known_sites": ["tests/data/database.vcf.gz"],
+  "reference":"tests/data/reference/ref.fa",
+  "dbsnp": "tests/data/reference/database.vcf.gz",
+  "known_sites": ["tests/data/reference/database.vcf.gz"],
   "scatter_size": 1000
 }

tests/data/config/sample_config_two_known_vcf.json

@@ -3,13 +3,13 @@
     "micro": {
       "libraries": {
         "lib_01": {
-        "R1": "tests/data/micro_R1.fq.gz",
-        "R2": "tests/data/micro_R2.fq.gz"
+        "R1": "tests/data/fastq/micro_R1.fq.gz",
+        "R2": "tests/data/fastq/micro_R2.fq.gz"
         }
       }
     }
   },
-  "reference":"tests/data/ref.fa",
-  "dbsnp": "tests/data/database.vcf.gz",
-  "known_sites": ["tests/data/database.vcf.gz", "tests/data/database.vcf.gz"]
+  "reference":"tests/data/reference/ref.fa",
+  "dbsnp": "tests/data/reference/database.vcf.gz",
+  "known_sites": ["tests/data/reference/database.vcf.gz", "tests/data/reference/database.vcf.gz"]
 }

tests/data/config/sample_config_two_readgroup.json

+{
+  "samples": {
+    "micro": {
+      "libraries": {
+        "lib_01": {
+        "R1": "tests/data/fastq/micro_rg1_R1.fq.gz",
+        "R2": "tests/data/fastq/micro_rg1_R2.fq.gz"
+        },
+        "lib_02": {
+        "R1": "tests/data/fastq/micro_rg2_R1.fq.gz",
+        "R2": "tests/data/fastq/micro_rg2_R2.fq.gz"
+        }
+      }
+    }
+  },
+  "reference":"tests/data/reference/ref.fa",
+  "dbsnp": "tests/data/reference/database.vcf.gz",
+  "known_sites": ["tests/data/reference/database.vcf.gz"],
+  "bedfile": "tests/data/reference/full_chrM.bed"
+}

tests/data/README.md → tests/data/fastq/README.md

@@ -9,3 +9,9 @@ fastq.
 The `ref.fa` fasta file contains human mitochondrial genome (NC_012920).
 
 The `database.vcf.gz` VCF file contains a single artificial variant. 
+
+The `micro_rg1_R1.fq.gz` and `micro_rg1_R2.fq.gz` were created by taking the
+first 15440 lines from `micro_R1.fq.gz` and `micro_R2.fq.gz, respectively.
+
+The `micro_rg2_R1.fq.gz` and `micro_rg2_R2.fq.gz` were created by taking the
+last 15440 lines from `micro_R1.fq.gz` and `micro_R2.fq.gz, respectively.