From d3c90c751696a089e85538825639bc2be131b4b4 Mon Sep 17 00:00:00 2001
From: Brian Pardy <brian.pardy@gmail.com>
Date: Tue, 25 Feb 2020 08:42:58 -0500
Subject: [PATCH 1/6] Add normalize_gisaid_fasta.sh for issue #53

---
 scripts/normalize_gisaid_fasta.sh | 35 +++++++++++++++++++++++++++++++
 1 file changed, 35 insertions(+)
 create mode 100755 scripts/normalize_gisaid_fasta.sh

diff --git a/scripts/normalize_gisaid_fasta.sh b/scripts/normalize_gisaid_fasta.sh
new file mode 100755
index 000000000..7a29f4ad5
--- /dev/null
+++ b/scripts/normalize_gisaid_fasta.sh
@@ -0,0 +1,35 @@
+#!/usr/bin/env bash
+set -e 
+GISAID_SARSCOV2_IN=$1
+GISAID_SARSCOV2_OUT=$2
+MIN_LENGTH=$3
+
+if [[ ! -r "$GISAID_SARSCOV2_IN" ]]
+then
+	echo "$0: input $GISAID_SARSCOV2_IN not found"
+	exit 1
+fi
+
+if [[ -z "$MIN_LENGTH" ]]
+then
+	echo "Using default minimum length of 15000"
+	MIN_LENGTH=15000
+fi
+
+echo "Normalizing GISAID file $GISAID_SARSCOV2_IN to $GISAID_SARSCOV2_OUT (min length $MIN_LENGTH)"
+
+# Remove leading 'BetaCoV' and 'BetaCov' from sequence names
+# Remove embedded spaces in sequence names (Hong Kong sequences)
+# Remove trailing |EPI_ISL_id|datestamp from sequence names
+# Remove sequences shorter than minimum length
+# Eliminate duplicate sequences (keep only the first seen)
+
+cat $GISAID_SARSCOV2_IN | 
+	sed 's/^>BetaCoV\//>/gi' |	# remove leading BetaCo[vV]
+	sed 's/ //g' |			# remove embedded spaces
+	sed 's/|.*$//' | 		# remove trailing metadata
+	awk "BEGIN{RS=\">\";FS=\"\n\"}length>$MIN_LENGTH{print \">\"\$0}" |	# remove short seqs
+	awk 'BEGIN{RS=">";FS="\n"}!x[$1]++{print ">"$0}' | 	# remove duplicates
+	grep -v '^>*$' > $GISAID_SARSCOV2_OUT
+
+exit 0

From 4a123001469e15103c4db36457f72e956bbea663 Mon Sep 17 00:00:00 2001
From: Brian Pardy <brian.pardy@gmail.com>
Date: Wed, 26 Feb 2020 08:24:06 -0500
Subject: [PATCH 2/6] Useless use of cat reported by Annatar removed

---
 scripts/normalize_gisaid_fasta.sh | 8 ++++----
 1 file changed, 4 insertions(+), 4 deletions(-)

diff --git a/scripts/normalize_gisaid_fasta.sh b/scripts/normalize_gisaid_fasta.sh
index 7a29f4ad5..8d4249214 100755
--- a/scripts/normalize_gisaid_fasta.sh
+++ b/scripts/normalize_gisaid_fasta.sh
@@ -24,10 +24,10 @@ echo "Normalizing GISAID file $GISAID_SARSCOV2_IN to $GISAID_SARSCOV2_OUT (min l
 # Remove sequences shorter than minimum length
 # Eliminate duplicate sequences (keep only the first seen)
 
-cat $GISAID_SARSCOV2_IN | 
-	sed 's/^>BetaCoV\//>/gi' |	# remove leading BetaCo[vV]
-	sed 's/ //g' |			# remove embedded spaces
-	sed 's/|.*$//' | 		# remove trailing metadata
+#cat $GISAID_SARSCOV2_IN | 
+	sed 's/^>BetaCoV\//>/gi' $GISAID_SARSCOV2_IN |	# remove leading BetaCo[vV]
+	sed 's/ //g' |					# remove embedded spaces
+	sed 's/|.*$//' | 				# remove trailing metadata
 	awk "BEGIN{RS=\">\";FS=\"\n\"}length>$MIN_LENGTH{print \">\"\$0}" |	# remove short seqs
 	awk 'BEGIN{RS=">";FS="\n"}!x[$1]++{print ">"$0}' | 	# remove duplicates
 	grep -v '^>*$' > $GISAID_SARSCOV2_OUT

From bc85d4b91dba2da69ffedf36853ac53011b92681 Mon Sep 17 00:00:00 2001
From: Brian Pardy <brian.pardy@gmail.com>
Date: Wed, 26 Feb 2020 12:02:46 -0500
Subject: [PATCH 3/6] Add gisaid rule to Snakefile

---
 Snakefile | 16 ++++++++++++++++
 1 file changed, 16 insertions(+)

diff --git a/Snakefile b/Snakefile
index 5f8bb0ac3..365427d58 100644
--- a/Snakefile
+++ b/Snakefile
@@ -420,3 +420,19 @@ rule clean:
         "auspice"
     shell:
         "rm -rfv {params}"
+
+rule gisaid:
+    message: "Normalizing GISAID download"
+    input:
+        gisaid_fasta = "data/gisaid_cov2020_sequences.fasta"
+    output:
+        sequences = "data/sequences.fasta"
+    params:
+        min_length = 15000
+    shell:
+        """
+	if [[ ! -f "data/sequences.fasta" && -f "data/gisaid_cov2020_sequences.fasta" ]]
+	then
+		scripts/normalize_gisaid_fasta.sh data/gisaid_cov2020_sequences.fasta data/sequences.fasta {params.min_length}
+	fi
+        """

From 1cf4613c80f99aed6ae21ba8c2215d86a03d2859 Mon Sep 17 00:00:00 2001
From: Brian Pardy <brian.pardy@gmail.com>
Date: Sun, 8 Mar 2020 13:38:57 -0400
Subject: [PATCH 4/6] Support GISAID renaming to hCoV-19

---
 scripts/normalize_gisaid_fasta.sh | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/scripts/normalize_gisaid_fasta.sh b/scripts/normalize_gisaid_fasta.sh
index 8d4249214..4668daacd 100755
--- a/scripts/normalize_gisaid_fasta.sh
+++ b/scripts/normalize_gisaid_fasta.sh
@@ -25,7 +25,7 @@ echo "Normalizing GISAID file $GISAID_SARSCOV2_IN to $GISAID_SARSCOV2_OUT (min l
 # Eliminate duplicate sequences (keep only the first seen)
 
 #cat $GISAID_SARSCOV2_IN | 
-	sed 's/^>BetaCoV\//>/gi' $GISAID_SARSCOV2_IN |	# remove leading BetaCo[vV]
+	sed 's/^>hCoV-19\//>/gi' $GISAID_SARSCOV2_IN |	# remove leading BetaCo[vV]
 	sed 's/ //g' |					# remove embedded spaces
 	sed 's/|.*$//' | 				# remove trailing metadata
 	awk "BEGIN{RS=\">\";FS=\"\n\"}length>$MIN_LENGTH{print \">\"\$0}" |	# remove short seqs

From 4f6c225fa0068cef7a012dd57bfbd158090a1dd2 Mon Sep 17 00:00:00 2001
From: Trevor Bedford <trevor@bedford.io>
Date: Sat, 14 Mar 2020 18:02:49 -0700
Subject: [PATCH 5/6] Fix typo in normalize_gisaid_fasta.sh

---
 scripts/normalize_gisaid_fasta.sh | 6 +++---
 1 file changed, 3 insertions(+), 3 deletions(-)

diff --git a/scripts/normalize_gisaid_fasta.sh b/scripts/normalize_gisaid_fasta.sh
index 4668daacd..0f8271662 100755
--- a/scripts/normalize_gisaid_fasta.sh
+++ b/scripts/normalize_gisaid_fasta.sh
@@ -1,5 +1,5 @@
 #!/usr/bin/env bash
-set -e 
+set -e
 GISAID_SARSCOV2_IN=$1
 GISAID_SARSCOV2_OUT=$2
 MIN_LENGTH=$3
@@ -24,8 +24,8 @@ echo "Normalizing GISAID file $GISAID_SARSCOV2_IN to $GISAID_SARSCOV2_OUT (min l
 # Remove sequences shorter than minimum length
 # Eliminate duplicate sequences (keep only the first seen)
 
-#cat $GISAID_SARSCOV2_IN | 
-	sed 's/^>hCoV-19\//>/gi' $GISAID_SARSCOV2_IN |	# remove leading BetaCo[vV]
+#cat $GISAID_SARSCOV2_IN |
+	sed 's/^>hCoV-19\//>/g' $GISAID_SARSCOV2_IN |	# remove leading BetaCo[vV]
 	sed 's/ //g' |					# remove embedded spaces
 	sed 's/|.*$//' | 				# remove trailing metadata
 	awk "BEGIN{RS=\">\";FS=\"\n\"}length>$MIN_LENGTH{print \">\"\$0}" |	# remove short seqs

From a869ac3873f61150fba3882795cb0706c7cdf171 Mon Sep 17 00:00:00 2001
From: Trevor Bedford <trevor@bedford.io>
Date: Sat, 14 Mar 2020 18:14:18 -0700
Subject: [PATCH 6/6] Remove gisaid rule from snakefile

Asking snakemake for "data/sequences.fasta" was throwing an error due to ambiguous DAG. This file can be created by either rule download or rule gisaid. There might be a better solution here, but in order to merge, I'm just going to remove from snakefile.
---
 Snakefile | 16 ----------------
 1 file changed, 16 deletions(-)

diff --git a/Snakefile b/Snakefile
index 365427d58..5f8bb0ac3 100644
--- a/Snakefile
+++ b/Snakefile
@@ -420,19 +420,3 @@ rule clean:
         "auspice"
     shell:
         "rm -rfv {params}"
-
-rule gisaid:
-    message: "Normalizing GISAID download"
-    input:
-        gisaid_fasta = "data/gisaid_cov2020_sequences.fasta"
-    output:
-        sequences = "data/sequences.fasta"
-    params:
-        min_length = 15000
-    shell:
-        """
-	if [[ ! -f "data/sequences.fasta" && -f "data/gisaid_cov2020_sequences.fasta" ]]
-	then
-		scripts/normalize_gisaid_fasta.sh data/gisaid_cov2020_sequences.fasta data/sequences.fasta {params.min_length}
-	fi
-        """