From 4ccee1a7deec539c47259a971e2aac7ff0d02556 Mon Sep 17 00:00:00 2001
From: Phoenix Logan <plogan@chanzuckerberg.com>
Date: Mon, 6 May 2024 09:33:11 -0700
Subject: [PATCH] [CZID-9457] - update PHAGE_FAMILIES_NAMES (#352)

* update PHAGE_FAMILIES_NAMES

* add more phage families

---------

Co-authored-by: phoenixAja <plogan@chanzuckerberg>
---
 .../index-generation/generate_lineage_csvs.py | 180 +++++++++++++++---
 1 file changed, 153 insertions(+), 27 deletions(-)

diff --git a/workflows/index-generation/generate_lineage_csvs.py b/workflows/index-generation/generate_lineage_csvs.py
index 1e7cda26e..a572c503e 100644
--- a/workflows/index-generation/generate_lineage_csvs.py
+++ b/workflows/index-generation/generate_lineage_csvs.py
@@ -43,33 +43,159 @@
 # We label as 'phage' all of the prokaryotic (bacterial and archaeal) virus families
 # listed here: https://en.wikipedia.org/wiki/Bacteriophage
 PHAGE_FAMILIES_NAMES = {
-    "Myoviridae",
-    "Siphoviridae",
-    "Podoviridae",
-    "Lipothrixviridae",
-    "Rudiviridae",
-    "Ampullaviridae",
-    "Bicaudaviridae",
-    "Clavaviridae",
-    "Corticoviridae",
-    "Cystoviridae",
-    "Fuselloviridae",
-    "Globuloviridae",
-    "Guttaviridae",
-    "Inoviridae",
-    "Leviviridae",
-    "Microviridae",
-    "Plasmaviridae",
-    "Tectiviridae",
-    "Turriviridae",
-    "Ackermannviridae",
-    "Sphaerolipoviridae",
-    "Pleolipoviridae",
-    "Finnlakeviridae",
-    "Portogloboviridae",
-    "Spiraviridae",
-    "Tristromaviridae",
-}
+    'Ackermannviridae',
+    'Aggregaviridae',
+    'Ahpuchviridae',
+    'Aliceevansviridae',
+    'Ampullaviridae',
+    'Anaerodiviridae',
+    'Andrewesvirinae',
+    'Aoguangviridae',
+    'Arenbergviridae',
+    'Armatusviridae',
+    'Arquatrovirinae',
+    'Assiduviridae',
+    'Atkinsviridae',
+    'Autographiviridae',
+    'Autolykiviridae',
+    'Azeredovirinae',
+    'Bclasvirinae',
+    'Beephvirinae',
+    'Bicaudaviridae',
+    'Blumeviridae',
+    'Boydwoodruffvirinae',
+    'Bronfenbrennervirinae',
+    'Casjensviridae',
+    'Ceeclamvirinae',
+    'Chaseviridae',
+    'Chebruvirinae',
+    'Chimalliviridae',
+    'Clavaviridae',
+    'Clermontviridae',
+    'Corticoviridae',
+    'Crevaviridae',
+    'Cystoviridae',
+    'Dclasvirinae',
+    'Deejayvirinae',
+    'Demerecviridae',
+    'Dolichocephalovirinae',
+    'Drexlerviridae',
+    'Druskaviridae',
+    'Duinviridae',
+    'Duneviridae',
+    'Eekayvirinae',
+    'Ekchuahviridae',
+    'Eucampyvirinae',
+    'Fervensviridae',
+    'Fiersviridae',
+    'Finnlakeviridae',
+    'Forsetiviridae',
+    'Fredfastierviridae',
+    'Fuselloviridae',
+    'Gclasvirinae',
+    'Globuloviridae',
+    'Gochnauervirinae',
+    'Gorgonvirinae',
+    'Graaviviridae',
+    'Gracegardnervirinae',
+    'Grimontviridae',
+    'Guelinviridae',
+    'Guenliviridae',
+    'Guernseyvirinae',
+    'Gutmannvirinae',
+    'Guttaviridae',
+    'Hafunaviridae',
+    'Haloferuviridae',
+    'Halomagnusviridae',
+    'Halspiviridae',
+    'Helgolandviridae',
+    'Hendrixvirinae',
+    'Herelleviridae',
+    'Inoviridae',
+    'Intestiviridae',
+    'Kairosviridae',
+    'Kantovirinae',
+    'Kleczkowskaviridae',
+    'Konodaiviridae',
+    'Kyanoviridae',
+    'Langleyhallvirinae',
+    'Leisingerviridae',
+    'Leviviridae',
+    'Lipothrixviridae',
+    'Lutetiaviridae',
+    'Madisaviridae',
+    'Madridviridae',
+    'Matshushitaviridae',
+    'Matsushitaviridae',
+    'Mccleskeyvirinae',
+    'Mesyanzhinovviridae',
+    'Microviridae',
+    'Molycolviridae',
+    'Myoviridae',
+    'Naomviridae',
+    'Nclasvirinae',
+    'Nymbaxtervirinae',
+    'Orlajensenviridae',
+    'Ounavirinae',
+    'Pachyviridae',
+    'Paulinoviridae',
+    'Pclasvirinae',
+    'Peduoviridae',
+    'Pervagoviridae',
+    'Pigerviridae',
+    'Plasmaviridae',
+    'Plectroviridae',
+    'Pleolipoviridae',
+    'Podoviridae',
+    'Pootjesviridae',
+    'Portogloboviridae',
+    'Pungoviridae',
+    'Pyrstoviridae',
+    'Queuovirinae',
+    'Rountreeviridae',
+    'Rudiviridae',
+    'Ruthgordonvirinae',
+    'Saffermanviridae',
+    'Salasmaviridae',
+    'Saparoviridae',
+    'Schitoviridae',
+    'Sepvirinae',
+    'Shortaselviridae',
+    'Simuloviridae',
+    'Siphoviridae',
+    'Skryabinvirinae',
+    'Soleiviridae',
+    'Solspiviridae',
+    'Speroviridae',
+    'Sphaerolipoviridae',
+    'Spiraviridae',
+    'Stanwilliamsviridae',
+    'Steigviridae',
+    'Steitzviridae',
+    'Stephanstirmvirinae',
+    'Straboviridae',
+    'Suolaviridae',
+    'Suoliviridae',
+    'Tectiviridae',
+    'Thaspiviridae',
+    'Toyamaviridae',
+    'Trabyvirinae',
+    'Tristromaviridae',
+    'Turriviridae',
+    'Tybeckvirinae',
+    'Umezonoviridae',
+    'Ungulaviridae',
+    'Vequintavirinae',
+    'Verdandiviridae',
+    'Vertoviridae',
+    'Vilmaviridae',
+    'Weiservirinae',
+    'Winoviridae',
+    'Yangangviridae',
+    'Yanlukaviridae',
+    'Zierdtviridae',
+    'Zobellviridae',
+ }
 
 
 def generate_taxon_lineage_names(