serovar list feature unit tested and new tests written, changelog upd…

…ated

CHANGELOG.md

@@ -91,21 +91,24 @@ Removed the following entries
 9. Chichiri,"6,14,24","z4,z24",-,,H,TRUE,enterica
 10. II 4:a:z39,"1,4,12,[27]",a,z39,,B,FALSE,salamae
 
-Modified the following entries
-1. Sundsvall,"[1],6,14,[25]",z,"e,n,x",,H,FALSE,enterica --> Sundsvall,"6,14",z,"e,n,x",,H,FALSE,enterica 
-2. Finkenwerder,"[1],6,14,[25]",d,"1,5",,H,FALSE,enterica --> Finkenwerder,"6,14",d,"1,5",,H,FALSE,enterica
-3. Florida,"[1],6,14,[25]",d,"1,7",,H,FALSE,enterica --> Florida,"6,14",d,"1,7",,H,FALSE,enterica
-4. Charity,"[1],6,14,[25]",d,"e,n,x",,H,FALSE,enterica --> Charity,"6,14",d,"e,n,x",,H,FALSE,enterica
-5. Onderstepoort,"1,6,14,[25]","e,h","1,5",,H,FALSE,enterica --> Onderstepoort,"6,14","e,h","1,5",,H,FALSE,enterica
-6. Schalkwijk,"6,14,[24]",i,"e,n,z15",,H,FALSE,enterica --> Schalkwijk,"6,14",i,"e,n,z15",,H,FALSE,enterica
-7. Boecker,"[1],6,14,[25]","l,v","1,7",,H,FALSE,enterica --> Boecker,"6,14","l,v","1,7",,H,FALSE,enterica
-8. Carrau,"6,14,[24]",y,"1,7",,H,FALSE,enterica --> Carrau,"6,14",y,"1,7",,H,FALSE,enterica
-9. Uzaramo,"1,6,14,25","z4,z24",-,,H,TRUE,enterica --> Uzaramo,"6,14","z4,z24",-,,H,TRUE,enterica
-10. Poano,"[1],6,14,[25]",z,"l,z13,z28",,H,FALSE,enterica --> Poano,"6,14",z,"l,z13,z28",,H,FALSE,enterica
+The following entries were modified in the in the `O_antigen` field as such
+
+| Before | After |
+|--------|-------|
+|Sundsvall,"[1],6,14,[25]",z,"e,n,x",,H,FALSE,enterica|  Sundsvall,"6,14",z,"e,n,x",,H,FALSE,enterica |
+|Finkenwerder,"[1],6,14,[25]",d,"1,5",,H,FALSE,enterica | Finkenwerder,"6,14",d,"1,5",,H,FALSE,enterica |
+|Florida,"[1],6,14,[25]",d,"1,7",,H,FALSE,enterica | Florida,"6,14",d,"1,7",,H,FALSE,enterica |
+| Charity,"[1],6,14,[25]",d,"e,n,x",,H,FALSE,enterica | Charity,"6,14",d,"e,n,x",,H,FALSE,enterica |
+| Onderstepoort,"1,6,14,[25]","e,h","1,5",,H,FALSE,enterica | Onderstepoort,"6,14","e,h","1,5",,H,FALSE,enterica |
+| Schalkwijk,"6,14,[24]",i,"e,n,z15",,H,FALSE,enterica | Schalkwijk,"6,14",i,"e,n,z15",,H,FALSE,enterica |
+| Boecker,"[1],6,14,[25]","l,v","1,7",,H,FALSE,enterica |Boecker,"6,14","l,v","1,7",,H,FALSE,enterica |
+| Carrau,"6,14,[24]",y,"1,7",,H,FALSE,enterica | Carrau,"6,14",y,"1,7",,H,FALSE,enterica |
+| Uzaramo,"1,6,14,25","z4,z24",-,,H,TRUE,enterica | Uzaramo,"6,14","z4,z24",-,,H,TRUE,enterica |
+| Poano,"[1],6,14,[25]",z,"l,z13,z28",,H,FALSE,enterica |  Poano,"6,14",z,"l,z13,z28",,H,FALSE,enterica |
 
 ### New fields 
 - Added `antigenic_formula` field that aggregates the O, H1 and H2 antigen values in a single location for convenience
-- Added `--list-of-serovars` option allowing user to provide a single column text file listing all serovars of interest to match against the SISTR prediction. This could be useful for cases when only a certain list of serovars could be reported
+- Added `--list-of-serovars` option allowing user to provide a single column text file listing all serovars of interest to match against the SISTR prediction. The result will be reportedi in `serovar_in_list` field as `Y` or `N` if there is match or otherwise. This could be useful for cases when only a certain list of serovars could be reported
 
 # 1.1.1
 

sistr/data/serovar-list.txt

@@ -0,0 +1,282 @@
+Abaetetuba
+Aberdeen
+Abony
+Adelaide
+Adjame
+Agama
+Agbeni
+Ago
+Agona
+Alachua
+Albany
+Altona
+Amager
+Amsterdam
+Anatum
+Anderlecht
+Apapa
+Apeyeme
+Arechavaleta
+Augustenborg
+Baildon
+Bangkok
+Bareilly
+Barranquilla
+Beaudesert
+Berta
+Birkenhead
+Blockley
+Bovismorbificans
+Braenderup
+Brancaster
+Brandenburg
+Bredeney
+Butantan
+Buzu
+Canada
+Cannstatt
+Carmel
+Carno
+Carrau
+Cerro
+Charity
+Chester
+Chicago
+Coeln
+Coleypark
+Colindale
+Colobane
+Concord
+Corvallis
+Cotham
+Cubana
+Curacao
+Daytona
+Denver
+Derby
+Dublin
+Durban
+Ealing
+Eastbourne
+Eko
+Elisabethville
+Enteritidis
+Florida
+Fluntern
+Freetown
+Fresno
+Galiema
+Gallinarum
+Gaminara
+Gatuni
+Give
+Goldcoast
+Grumpensis
+Guinea
+Hadar
+Haifa
+Hartford
+Havana
+Heidelberg
+Herston
+Holcomb
+Hvittingfoss
+I 1,4,[5],12:b:-
+I 1,4,[5],12:d:-
+I 1,4,[5],12:i:-
+I 1,4,[5],12:r:-
+Ibadan
+Idikan
+II 42:b:e,n,x,z15
+II 42:r:-
+II 42:z:1,5
+II 47:b:e,n,x,z15
+II 48:d:z6
+II 50:b:z6
+II 58:d:z6
+II 58:l,z13,z28:z6
+IIIa 1,13,23:g,z51:-
+IIIa 18:z4,z23:-
+IIIa 40:z4,z23:-
+IIIa 41:z4,z23:-
+IIIa 44:z4,z23,z32:-
+IIIa 44:z4,z23:-
+IIIa 48:z4,z24:-
+IIIa 56:z4,z23,z32:-
+IIIb 16:z10:e,n,x,z15
+IIIb 38:k:z
+IIIb 38:k:z35
+IIIb 48:l,v:1,5,7
+IIIb 48:z52:z
+IIIb 50:k:z
+IIIb 50:r:z
+IIIb 53:z10:z35
+IIIb 60:r:e,n,x,z15
+IIIb 60:r:z
+IIIb 61:i:z53
+IIIb 61:k:1,5,7
+IIIb 61:k:z35
+IIIb 61:l,v:1,5,7
+IIIb 61:l,v:z35
+IIIb 61:r:z
+IIIb 61:z52:z53
+IIIb 65:k:z
+IIIb 65:k:z53
+IIIb 65:z10:e,n,x,z15
+Indiana
+Infantis
+Inganda
+Inverness
+Irumu
+Isangi
+Ituri
+IV 11:g,z51:-
+IV 16:z4,z32:-
+IV 40:z4,z24:-
+IV 43:z4,z23:-
+IV 44:z36,z38:-
+IV 44:z4,z23:-
+IV 44:z4,z32:-
+IV 45:g,z51:-
+IV 48:g,z51:-
+IV 48:z4,z32:-
+IV 50:g,z51:-
+IV 50:z4,z23:-
+Jangwani
+Javiana
+Johannesburg
+Kambole
+Kandla
+Kedougou
+Kentucky
+Kenya
+Kiambu
+Kimuenza
+Kingston
+Kirkee
+Kisarawe
+Kottbus
+Kumasi
+Lagos
+Lansing
+Larochelle
+Lattenkamp
+Lexington
+Lille
+Limete
+Litchfield
+Liverpool
+Livingstone
+Lomalinda
+Lome
+London
+Louga
+Lubbock
+Luciana
+Luke
+Malstatt
+Manchester
+Manhattan
+Marseille
+Matadi
+Matopeni
+Mbandaka
+Meleagridis
+Mgulani
+Miami
+Michigan
+Mikawasima
+Milwaukee
+Minnesota
+Mishmarhaemek
+Mississippi
+Molade
+Monschaui
+Montevideo
+Mountpleasant
+Muenchen
+Muenster
+Napoli
+Nchanga
+Newlands
+Newport
+Nigeria
+Norwich
+Nottingham
+Obogu
+Offa
+Ohio
+Onderstepoort
+Onireke
+Oranienburg
+Ordonez
+Orientalis
+Orion
+Oslo
+Ouakam
+Overschie
+Panama
+Paratyphi A
+Paratyphi B
+Paratyphi B var. Java
+Pensacola
+Perth
+Plymouth
+Poano
+Pomona
+Poona
+Potsdam
+Putten
+Reading
+Redlands
+Richmond
+Ridge
+Rissen
+Roodepoort
+Rubislaw
+Saintpaul
+Sandiego
+Sanjuan
+Schwarzengrund
+Sendai
+Senegal
+Senftenberg
+Shubra
+Singapore
+Sinstorf
+Soerenga
+Soumbedioune
+Stanley
+Stanleyville
+Stockholm
+Stourbridge
+Suelldorf
+Sundsvall
+Szentes
+Tafo
+Takoradi
+Tallahassee
+Tamberma
+Tanger
+Tarshyne
+Teddington
+Tees
+Telaviv
+Telelkebir
+Tennessee
+Thompson
+Tudu
+Typhi
+Typhimurium
+Uganda
+Umbilo
+Urbana
+Uzaramo
+Virchow
+Vitkin
+Wandsworth
+Wangata
+Weltevreden
+Wien
+Worthington
+Zaiman

sistr/sistr_cmd.py

@@ -93,9 +93,9 @@ def init_parser():
                         type=int,
                         default=1,
                         help='Number of parallel threads to run sistr_cmd analysis.')
-    parser.add_argument('-l', '--list-of-serovars',
-                        type=str, required=False,
-                        help='A path to a single column text file containing list of serovar(s) to check serovar prediction against. Report predicted serovar is Y (present) and N (absent) in the list')
+    parser.add_argument('-l', '--list-of-serovars', nargs='?',
+                        required=False, const=os.path.join(os.path.dirname(os.path.abspath(__file__)), "data/serovar-list.txt"),
+                        help='A path to a single column text file containing list of serovar(s) to check serovar prediction against. Report predicted serovar in "serovar_in_list" field as  Y (present) and N (absent) in the list. The default list will be used if not file specified.')
     parser.add_argument('-v',
                         '--verbose',
                         action='count',
@@ -256,7 +256,7 @@ def sistr_predict(input_fasta, genome_name, tmp_dir, keep_tmp, args):
         if os.path.exists(args.list_of_serovars):
             with open(args.list_of_serovars) as fp:
                 serovars_selected_list = [l.rstrip() for l in fp.readlines()]
-            logging.info(f"Selected serovars list to check predictions against from {args.list_of_serovars} is {serovars_selected_list}")
+            logging.info(f"Selected serovars list to check predictions against from {args.list_of_serovars} containing {len(serovars_selected_list)} serovars")
         else:
             logging.warning(f"File {args.list_of_serovars} does not exist in path specified. Would not check against list of provided serovars ...")      
 

tests/test_serotyping.py

@@ -0,0 +1,23 @@
+import sys, os, json
+from sistr.sistr_cmd import main
+
+
+
+def test_serotyping(fasta_path):
+    sys.argv[1:] = ["-i",fasta_path, "test", "--more-results", "--run-mash", "--qc", "-l", "-vvv", "-o", "sistr_test_results"]
+    main()
+    assert os.path.exists("sistr_test_results.json"), f"Results SISTR file for {fasta_path} not found"
+    with open("sistr_test_results.json") as fp:
+        sistr_json_results = json.load(fp)[0]
+    assert  sistr_json_results['antigenic_formula'] == "58:l,z13,z28:z6"
+    assert  sistr_json_results['cgmlst_matching_alleles'] == 330
+    assert  sistr_json_results['cgmlst_found_loci'] == 330
+    assert  sistr_json_results['serovar_in_list'] == "Y"
+    assert  sistr_json_results['serovar'] == "II 58:l,z13,z28:z6"
+
+def test_noserovarlist_file(fasta_path):
+    sys.argv[1:] = ["-i",fasta_path, "test", "--more-results", "--run-mash", "--qc", "-l", "no_file_exists.txt", "-vvv", "-o", "sistr_test_results"]
+    main() 
+    with open("sistr_test_results.json") as fp:
+        sistr_json_results = json.load(fp)[0]
+    assert 'serovar_in_list' not in sistr_json_results.keys()