New flu subtypes (H5NX, H7NX, H9NX) (#620)

* Add new subtypes: H5NX, H7NX, H9NX. Also, add putative signal peptides for B subtypes (B-vic, B-yam) * Update flu ingest structure * Use configurable min_date for global_seq_data analysis * Shorten sequence mutation table name to accomodate longer partition names * Add new subtypes to flu config files
vector-engineering · May 1, 2023 · 4598e5a · 4598e5a
1 parent c91fdb3
commit 4598e5a
Show file tree

Hide file tree

Showing 69 changed files with 1,613 additions and 117 deletions.
diff --git a/config/config_flu_genbank.yaml b/config/config_flu_genbank.yaml
@@ -80,6 +80,9 @@ report_group_references:
   B-yam: B-Phuket-3073-2013
   H1N1: A-Wisconsin-67-2022
   H3N2: A-Darwin-6-2021
+  H5NX: A-Goose-Guangdong-1-96
+  H7NX: A-Shanghai-02-2013
+  H9NX: A-Hong-Kong-1073-99
 
 # Surveillance plot options
 # see: workflow_main/scripts/surveillance.py
@@ -95,6 +98,9 @@ surv_group_references:
   B-yam: B-Phuket-3073-2013
   H1N1: A-Wisconsin-67-2022
   H3N2: A-Darwin-6-2021
+  H5NX: A-Goose-Guangdong-1-96
+  H7NX: A-Shanghai-02-2013
+  H9NX: A-Hong-Kong-1073-99
 
 # ---------------
 #    DATABASE
@@ -134,6 +140,9 @@ default_references:
   B-yam: B-Phuket-3073-2013
   H1N1: A-Wisconsin-67-2022
   H3N2: A-Darwin-6-2021
+  H5NX: A-Goose-Guangdong-1-96
+  H7NX: A-Shanghai-02-2013
+  H9NX: A-Hong-Kong-1073-99
 
 # Home page
 show_home_banner: false

diff --git a/config/config_flu_gisaid.yaml b/config/config_flu_gisaid.yaml
@@ -84,6 +84,9 @@ report_group_references:
   B-yam: B-Phuket-3073-2013
   H1N1: A-Wisconsin-67-2022
   H3N2: A-Darwin-6-2021
+  H5NX: A-Goose-Guangdong-1-96
+  H7NX: A-Shanghai-02-2013
+  H9NX: A-Hong-Kong-1073-99
 
 # Surveillance plot options
 # see: workflow_main/scripts/surveillance.py
@@ -99,6 +102,9 @@ surv_group_references:
   B-yam: B-Phuket-3073-2013
   H1N1: A-Wisconsin-67-2022
   H3N2: A-Darwin-6-2021
+  H5NX: A-Goose-Guangdong-1-96
+  H7NX: A-Shanghai-02-2013
+  H9NX: A-Hong-Kong-1073-99
 
 # ---------------
 #    DATABASE
@@ -139,6 +145,9 @@ default_references:
   B-yam: B-Phuket-3073-2013
   H1N1: A-Wisconsin-67-2022
   H3N2: A-Darwin-6-2021
+  H5NX: A-Goose-Guangdong-1-96
+  H7NX: A-Shanghai-02-2013
+  H9NX: A-Hong-Kong-1073-99
 
 # Home page
 show_home_banner: false

diff --git a/config/config_flu_gisaid_dev.yaml b/config/config_flu_gisaid_dev.yaml
@@ -84,6 +84,9 @@ report_group_references:
   B-yam: B-Phuket-3073-2013
   H1N1: A-Wisconsin-67-2022
   H3N2: A-Darwin-6-2021
+  H5NX: A-Goose-Guangdong-1-96
+  H7NX: A-Shanghai-02-2013
+  H9NX: A-Hong-Kong-1073-99
 
 # Surveillance plot options
 # see: workflow_main/scripts/surveillance.py
@@ -99,6 +102,9 @@ surv_group_references:
   B-yam: B-Phuket-3073-2013
   H1N1: A-Wisconsin-67-2022
   H3N2: A-Darwin-6-2021
+  H5NX: A-Goose-Guangdong-1-96
+  H7NX: A-Shanghai-02-2013
+  H9NX: A-Hong-Kong-1073-99
 
 # ---------------
 #    DATABASE
@@ -139,6 +145,9 @@ default_references:
   B-yam: B-Phuket-3073-2013
   H1N1: A-Wisconsin-67-2022
   H3N2: A-Darwin-6-2021
+  H5NX: A-Goose-Guangdong-1-96
+  H7NX: A-Shanghai-02-2013
+  H9NX: A-Hong-Kong-1073-99
 
 # Home page
 show_home_banner: false

diff --git a/services/server/cg_server/db_seed/seed.py b/services/server/cg_server/db_seed/seed.py
@@ -480,7 +480,7 @@ def seed_database(conn, schema="public"):
                 # Clean up the reference name as a SQL ident - no dots
                 reference_name_sql = reference_name.replace(".", "_")
 
-                reference_partition_name = f"{table_name}_{reference_name_sql}"
+                reference_partition_name = f"seqmut_{mutation_field}_{reference_name_sql}"
 
                 # Create reference partition
                 cur.execute(
@@ -511,7 +511,7 @@ def seed_database(conn, schema="public"):
                         """
                         ).format(
                             date_partition_name=sql.Identifier(
-                                f"{table_name}_{reference_name_sql}_{i}"
+                                f"seqmut_{mutation_field}_{reference_name_sql}_{i}"
                             ),
                             reference_partition_name=sql.Identifier(
                                 reference_partition_name

diff --git a/static_data/flu/alignments/B-vic_4_FLBHAAA.dna b/static_data/flu/alignments/B-vic_4_FLBHAAA.dna
diff --git a/static_data/flu/alignments/B-yam_4_MK715607.dna b/static_data/flu/alignments/B-yam_4_MK715607.dna
diff --git a/static_data/flu/alignments/H5N1_1_NC_007357.dna b/static_data/flu/alignments/H5N1_1_NC_007357.dna
diff --git a/static_data/flu/alignments/H5N1_2_NC_007358.dna b/static_data/flu/alignments/H5N1_2_NC_007358.dna
diff --git a/static_data/flu/alignments/H5N1_3_NC_007359.dna b/static_data/flu/alignments/H5N1_3_NC_007359.dna
diff --git a/static_data/flu/alignments/H5N1_4_NC_007362.dna b/static_data/flu/alignments/H5N1_4_NC_007362.dna
diff --git a/static_data/flu/alignments/H5N1_5_NC_007360.dna b/static_data/flu/alignments/H5N1_5_NC_007360.dna
diff --git a/static_data/flu/alignments/H5N1_6_NC_007361.dna b/static_data/flu/alignments/H5N1_6_NC_007361.dna
diff --git a/static_data/flu/alignments/H5N1_7_NC_007363.dna b/static_data/flu/alignments/H5N1_7_NC_007363.dna
diff --git a/static_data/flu/alignments/H5N1_8_NC_007364.dna b/static_data/flu/alignments/H5N1_8_NC_007364.dna
diff --git a/static_data/flu/alignments/H7N2_1_NC_026422.dna b/static_data/flu/alignments/H7N2_1_NC_026422.dna
diff --git a/static_data/flu/alignments/H7N2_2_NC_026423.dna b/static_data/flu/alignments/H7N2_2_NC_026423.dna
diff --git a/static_data/flu/alignments/H7N2_3_NC_026424.dna b/static_data/flu/alignments/H7N2_3_NC_026424.dna
diff --git a/static_data/flu/alignments/H7N2_4_NC_026425.dna b/static_data/flu/alignments/H7N2_4_NC_026425.dna
diff --git a/static_data/flu/alignments/H7N2_5_NC_026426.dna b/static_data/flu/alignments/H7N2_5_NC_026426.dna
diff --git a/static_data/flu/alignments/H7N2_6_NC_026429.dna b/static_data/flu/alignments/H7N2_6_NC_026429.dna
diff --git a/static_data/flu/alignments/H7N2_7_NC_026427.dna b/static_data/flu/alignments/H7N2_7_NC_026427.dna
diff --git a/static_data/flu/alignments/H7N2_8_NC_026428.dna b/static_data/flu/alignments/H7N2_8_NC_026428.dna
diff --git a/static_data/flu/alignments/H9N2_1_NC_004910.dna b/static_data/flu/alignments/H9N2_1_NC_004910.dna
diff --git a/static_data/flu/alignments/H9N2_2_NC_004911.dna b/static_data/flu/alignments/H9N2_2_NC_004911.dna
diff --git a/static_data/flu/alignments/H9N2_3_NC_004912.dna b/static_data/flu/alignments/H9N2_3_NC_004912.dna
diff --git a/static_data/flu/alignments/H9N2_4_NC_004908.dna b/static_data/flu/alignments/H9N2_4_NC_004908.dna
diff --git a/static_data/flu/alignments/H9N2_5_NC_004905.dna b/static_data/flu/alignments/H9N2_5_NC_004905.dna
diff --git a/static_data/flu/alignments/H9N2_6_NC_004909.dna b/static_data/flu/alignments/H9N2_6_NC_004909.dna
diff --git a/static_data/flu/alignments/H9N2_7_NC_004907.dna b/static_data/flu/alignments/H9N2_7_NC_004907.dna
diff --git a/static_data/flu/alignments/H9N2_8_NC_004906.dna b/static_data/flu/alignments/H9N2_8_NC_004906.dna
diff --git a/static_data/flu/alignments/HA_all.praln b/static_data/flu/alignments/HA_all.praln
diff --git a/static_data/flu/genes.csv b/static_data/flu/genes.csv
@@ -130,7 +130,7 @@ H5NX,A-Goose-Guangdong-1-96,NS1,8,15..707,1,0,[],
 H7NX,A-Shanghai-02-2013,PB2,1,1..2280,1,0,[],
 H7NX,A-Shanghai-02-2013,PB1,2,1..2274,1,0,[],
 H7NX,A-Shanghai-02-2013,PA,3,1..2151,1,0,[],
-H7NX,A-Shanghai-02-2013,HA,4,1..1683,1,16,"[{""name"": ""signal peptide"", ""ranges"": [[1, 18]]}, {""name"": ""HA1"", ""ranges"": [[19, 339]]}, {""name"": ""HA2"", ""ranges"": [[340, 560]]}]",
+H7NX,A-Shanghai-02-2013,HA,4,1..1683,1,18,"[{""name"": ""signal peptide"", ""ranges"": [[1, 18]]}, {""name"": ""HA1"", ""ranges"": [[19, 339]]}, {""name"": ""HA2"", ""ranges"": [[340, 560]]}]",
 H7NX,A-Shanghai-02-2013,NP,5,1..1497,1,0,[],
 H7NX,A-Shanghai-02-2013,NA,6,1..1398,1,0,[],
 H7NX,A-Shanghai-02-2013,M1,7,1..759,1,0,[],
@@ -140,55 +140,55 @@ H7NX,A-Shanghai-02-2013,NS1,8,1..654,1,0,[],
 H9NX,A-Hong-Kong-1073-99,PB2,1,28..2307,1,0,[],
 H9NX,A-Hong-Kong-1073-99,PB1,2,24..2300,1,0,[],
 H9NX,A-Hong-Kong-1073-99,PA,3,21..2171,1,0,[],
-H9NX,A-Hong-Kong-1073-99,HA,4,1..1714,1,16,"[{""name"": ""signal peptide"", ""ranges"": [[1, 18]]}, {""name"": ""HA1"", ""ranges"": [[19, 338]]}, {""name"": ""HA2"", ""ranges"": [[339, 560]]}]",
+H9NX,A-Hong-Kong-1073-99,HA,4,1..1714,1,18,"[{""name"": ""signal peptide"", ""ranges"": [[1, 18]]}, {""name"": ""HA1"", ""ranges"": [[19, 338]]}, {""name"": ""HA2"", ""ranges"": [[339, 560]]}]",
 H9NX,A-Hong-Kong-1073-99,NP,5,40..1536,1,0,[],
 H9NX,A-Hong-Kong-1073-99,NA,6,1..1404,1,0,[],
 H9NX,A-Hong-Kong-1073-99,M1,7,33..791,1,0,[],
 H9NX,A-Hong-Kong-1073-99,M2,7,33..59;748..1014,1,0,[],
 H9NX,A-Hong-Kong-1073-99,NEP,8,27..56;529..864,1,0,[],
 H9NX,A-Hong-Kong-1073-99,NS1,8,27..719,1,0,[],
-B-yam,B-Massachusetts-02-2012,HA,4,1..1755,1,0,[],
+B-yam,B-Massachusetts-02-2012,HA,4,1..1755,1,15,"[{""name"": ""signal peptide"", ""ranges"": [[1, 15]]}, {""name"": ""HA1"", ""ranges"": [[16, 361]]}, {""name"": ""HA2"", ""ranges"": [[362, 584]]}]",
 B-yam,B-Massachusetts-02-2012,NA,6,1..1401,1,0,[],
 B-yam,B-Massachusetts-02-2012,NB,6,3..296,1,0,[],
 B-yam,B-Massachusetts-02-2012,NEP,8,1..33;689..1024,1,0,[],
 B-yam,B-Massachusetts-02-2012,NS1,8,1..846,1,0,[],
 B-yam,B-Phuket-3073-2013,PB2,1,1..2313,1,0,[],
 B-yam,B-Phuket-3073-2013,PB1,2,1..2259,1,0,[],
 B-yam,B-Phuket-3073-2013,PA,3,1..2181,1,0,[],
-B-yam,B-Phuket-3073-2013,HA,4,1..1755,1,0,[],
+B-yam,B-Phuket-3073-2013,HA,4,1..1755,1,15,"[{""name"": ""signal peptide"", ""ranges"": [[1, 15]]}, {""name"": ""HA1"", ""ranges"": [[16, 361]]}, {""name"": ""HA2"", ""ranges"": [[362, 584]]}]",
 B-yam,B-Phuket-3073-2013,NP,5,1..1683,1,0,[],
 B-yam,B-Phuket-3073-2013,NA,6,34..1434,1,0,[],
 B-yam,B-Phuket-3073-2013,NB,6,27..329,1,0,[],
 B-yam,B-Phuket-3073-2013,M,7,1..747,1,0,[],
 B-yam,B-Phuket-3073-2013,BM2,7,747..1076,1,0,[],
 B-yam,B-Phuket-3073-2013,NEP,8,1..33;689..1024,1,0,[],
 B-yam,B-Phuket-3073-2013,NS1,8,1..846,1,0,[],
-B-yam,B-Wisconsin-01-2010,HA,4,1..1755,1,0,[],
+B-yam,B-Wisconsin-01-2010,HA,4,1..1755,1,15,"[{""name"": ""signal peptide"", ""ranges"": [[1, 15]]}, {""name"": ""HA1"", ""ranges"": [[16, 361]]}, {""name"": ""HA2"", ""ranges"": [[362, 584]]}]",
 B-yam,B-Wisconsin-01-2010,NA,6,1..1401,1,0,[],
 B-yam,B-Wisconsin-01-2010,NB,6,3..296,1,0,[],
 B-yam,B-Wisconsin-01-2010,NEP,8,1..33;689..1024,1,0,[],
 B-yam,B-Wisconsin-01-2010,NS1,8,1..846,1,0,[],
 B-vic,B-Austria-1359417-2021,PB2,1,1..2313,1,0,[],
 B-vic,B-Austria-1359417-2021,PB1,2,1..2259,1,0,[],
 B-vic,B-Austria-1359417-2021,PA,3,1..2181,1,0,[],
-B-vic,B-Austria-1359417-2021,HA,4,1..1749,1,0,[],
+B-vic,B-Austria-1359417-2021,HA,4,1..1749,1,15,"[{""name"": ""signal peptide"", ""ranges"": [[1, 15]]}, {""name"": ""HA1"", ""ranges"": [[16, 359]]}, {""name"": ""HA2"", ""ranges"": [[360, 582]]}]",
 B-vic,B-Austria-1359417-2021,NP,5,1..1683,1,0,[],
 B-vic,B-Austria-1359417-2021,NA,6,8..1408,1,0,[],
 B-vic,B-Austria-1359417-2021,NB,6,1..303,1,0,[],
 B-vic,B-Austria-1359417-2021,M,7,1..747,1,0,[],
 B-vic,B-Austria-1359417-2021,BM2,7,747..1076,1,0,[],
 B-vic,B-Austria-1359417-2021,NEP,8,1..36;692..1027,1,0,[],
 B-vic,B-Austria-1359417-2021,NS1,8,1..849,1,0,[],
-B-vic,B-Brisbane-60-2008,HA,4,34..1791,1,0,[],
-B-vic,B-Colorado-06-2017,HA,4,34..1785,1,0,[],
+B-vic,B-Brisbane-60-2008,HA,4,34..1791,1,15,"[{""name"": ""signal peptide"", ""ranges"": [[1, 15]]}, {""name"": ""HA1"", ""ranges"": [[16, 362]]}, {""name"": ""HA2"", ""ranges"": [[363, 585]]}]",
+B-vic,B-Colorado-06-2017,HA,4,34..1785,1,15,"[{""name"": ""signal peptide"", ""ranges"": [[1, 15]]}, {""name"": ""HA1"", ""ranges"": [[16, 360]]}, {""name"": ""HA2"", ""ranges"": [[361, 583]]}]",
 B-vic,B-Colorado-06-2017,NA,6,54..1454,1,0,[],
 B-vic,B-Colorado-06-2017,NB,6,47..349,1,0,[],
 B-vic,B-Colorado-06-2017,M,7,25..771,1,0,[],
 B-vic,B-Colorado-06-2017,BM2,7,771..1100,1,0,[],
 B-vic,B-Washington-02-2019,PB2,1,10..2322,1,0,[],
 B-vic,B-Washington-02-2019,PB1,2,8..2266,1,0,[],
 B-vic,B-Washington-02-2019,PA,3,16..2196,1,0,[],
-B-vic,B-Washington-02-2019,HA,4,20..1768,1,0,[],
+B-vic,B-Washington-02-2019,HA,4,20..1768,1,15,"[{""name"": ""signal peptide"", ""ranges"": [[1, 15]]}, {""name"": ""HA1"", ""ranges"": [[16, 359]]}, {""name"": ""HA2"", ""ranges"": [[360, 582]]}]",
 B-vic,B-Washington-02-2019,NP,5,47..1729,1,0,[],
 B-vic,B-Washington-02-2019,NA,6,40..1440,1,0,[],
 B-vic,B-Washington-02-2019,NB,6,33..335,1,0,[],