From 8fd9853a4c6e5b0eadfb1d33f2f6581cd817364d Mon Sep 17 00:00:00 2001
From: Daniel Cook <danielecook@gmail.com>
Date: Sat, 26 Oct 2019 19:34:17 +0100
Subject: [PATCH] fix tests and novaseq detection

---
 scripts/functional-tests.sh | 16 +++++++++++-----
 src/fq_meta.nim             | 14 +++++++-------
 2 files changed, 18 insertions(+), 12 deletions(-)

diff --git a/scripts/functional-tests.sh b/scripts/functional-tests.sh
index 5a9aa4f..27f77e8 100644
--- a/scripts/functional-tests.sh
+++ b/scripts/functional-tests.sh
@@ -19,7 +19,7 @@ assert_equal 999 "$(cat "$STDOUT_FILE" | jq '.QUAL')"
 assert_equal \"T\" "$(cat "$STDOUT_FILE" | jq '.REF')"
 
 run test_json_pretty sc json --pretty "${PARENT_DIR}/tests/data/test.vcf.gz" X:17276844-17276844
-assert_equal 13 "$(cat $STDOUT_FILE | wc -l)"
+assert_equal 13 $(cat $STDOUT_FILE | wc -l)
 
 # INFO
 run single_info_item sc json --info="DP" "${PARENT_DIR}/tests/data/test.vcf.gz" X:17276844-17276844
@@ -40,12 +40,12 @@ assert_equal 2094 "$(cat $STDOUT_FILE | jq '.FORMAT.DP|add')"
 #==========#
 
 run fq_dedup sc fq-dedup tests/fastq/dup.fq
-assert_equal 4 "$(cat $STDOUT_FILE | grep '@' | wc -l)"
-assert_equal 4 "$(cat $STDOUT_FILE | grep '@' | wc -l)"
+assert_equal 4 $(cat $STDOUT_FILE | grep '@' | wc -l)
+assert_equal 4 $(cat $STDOUT_FILE | grep '@' | wc -l)
 
 run fq_dedup_gz sc fq-dedup tests/fastq/dup.fq.gz
-assert_equal 4 "$(cat $STDOUT_FILE | grep '@' | wc -l)"
-assert_equal 4 "$(cat $STDOUT_FILE | grep '@' | wc -l)"
+assert_equal 4 $(cat $STDOUT_FILE | grep '@' | wc -l)
+assert_equal 4 $(cat $STDOUT_FILE | grep '@' | wc -l)
 
 #=========#
 # fq-meta #
@@ -112,3 +112,9 @@ assert_equal "$(cat $STDOUT_FILE | cut -f 2)" "HiSeqX"
 assert_equal "$(cat $STDOUT_FILE | cut -f 3)" "high:machine+flowcell"
 assert_equal "$(cat $STDOUT_FILE | cut -f 3 | cut -f 1 -d ':')" \
              "$(test_fq tests/fastq/illumina_hiseq_x.fq | egrep -o 'high')"
+
+# Novaseq
+run fq_meta_novaseq sc fq-meta tests/fastq/novaseq.fq
+assert_equal "$(cat $STDOUT_FILE | cut -f 2)" "NovaSeq"
+assert_equal "$(cat $STDOUT_FILE | cut -f 3)" "high:machine+flowcell"
+assert_equal "$(cat $STDOUT_FILE | cut -f 3 | cut -f 1 -d ':')"
\ No newline at end of file
diff --git a/src/fq_meta.nim b/src/fq_meta.nim
index 9c912f5..49dbf90 100644
--- a/src/fq_meta.nim
+++ b/src/fq_meta.nim
@@ -52,7 +52,7 @@ type
 let InstrumentIDs = @[Instrument(pattern: re"HWI-M[0-9]{4}$", sequencer: @["MiSeq"]),
                       Instrument(pattern: re"HWUSI", sequencer: @["GenomeAnalyzerIIx"]),
                       Instrument(pattern: re"M[0-9]{5}$", sequencer: @["MiSeq"]),
-                      Instrument(pattern: re"A[0-9]{6}$", sequencer: @["NovaSeq"]),
+                      Instrument(pattern: re"A[0-9]{5}$", sequencer: @["NovaSeq"]),
                       Instrument(pattern: re"HWI-C[0-9]{5}$", sequencer: @["HiSeq1500"]),
                       Instrument(pattern: re"C[0-9]{5}$", sequencer: @["HiSeq1500"]),
                       Instrument(pattern: re"HWI-D[0-9]{5}$", sequencer: @["HiSeq2500"]),
@@ -89,12 +89,12 @@ let FCIDs = @[Flowcell(pattern: re"AAXX$", sequencer: @["GenomeAnalyzer"], descr
               Flowcell(pattern: re"H[A-Z,0-9]{4}BGXY$", sequencer: @["NextSeq"], description: "High output flow cell"),
               Flowcell(pattern: re"H[A-Z,0-9]{4}BGX2$", sequencer: @["NextSeq"], description: "High output flow cell"),
               Flowcell(pattern: re"H[A-Z,0-9]{4}AFXX$", sequencer: @["NextSeq"], description: "Mid output flow cell"),
-              Flowcell(pattern: re"A[A-Z,0-9]{4}$", sequencer: @["MiSeq"], description: "MiSeq flow cell"),
-              Flowcell(pattern: re"B[A-Z,0-9]{4}$", sequencer: @["MiSeq"], description: "MiSeq flow cell"),
-              Flowcell(pattern: re"D[A-Z,0-9]{4}$", sequencer: @["MiSeq"], description: "MiSeq nano flow cell"),
-              Flowcell(pattern: re"G[A-Z,0-9]{4}$", sequencer: @["MiSeq"], description: "MiSeq micro flow cell"),
-              Flowcell(pattern: re"H[A-Z,0-9]{4}DMXX$", sequencer: @["NovaSeq"], description: "S2 flow cell")]
-
+              Flowcell(pattern: re"H[A-Z,0-9]{4}DMXX$", sequencer: @["NovaSeq"], description: "S2 flow cell"),
+              Flowcell(pattern: re"H[A-Z,0-9]{4}DSXX$", sequencer: @["NovaSeq"], description: "S2 flow cell"),
+              Flowcell(pattern: re"^A[A-Z,0-9]{4}$", sequencer: @["MiSeq"], description: "MiSeq flow cell"),
+              Flowcell(pattern: re"^B[A-Z,0-9]{4}$", sequencer: @["MiSeq"], description: "MiSeq flow cell"),
+              Flowcell(pattern: re"^D[A-Z,0-9]{4}$", sequencer: @["MiSeq"], description: "MiSeq nano flow cell"),
+              Flowcell(pattern: re"^G[A-Z,0-9]{4}$", sequencer: @["MiSeq"], description: "MiSeq micro flow cell")]
 
 proc qual_to_int(q_score: char): int =
     return qual.find(q_score)