Spaces:

EphAsad
/

BactKing

Sleeping

App Files Files Community

EphAsad commited on Dec 16, 2025

Commit

a1dc114

verified ·

1 Parent(s): 81668a9

train: update extended schema, aliases, signals from gold tests

Browse files

Files changed (2) hide show

models/genus_xgb.json +2 -2
models/genus_xgb_meta.json +282 -248

models/genus_xgb.json CHANGED Viewed

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:af676e86223286f6a32391219b3054ebe309199bbc31bf39f3248edc16114b9d
-size 22523681

 version https://git-lfs.github.com/spec/v1
+oid sha256:6530346e18bdb61e778f887fef7ca33e8e0b56e040ce95287c373073db726cfb
+size 34613851

models/genus_xgb_meta.json CHANGED Viewed

@@ -1,260 +1,294 @@
 {
   "genus_to_idx": {
-    "Weissella": 0,
-    "Vibrio": 1,
-    "Spiroplasma": 2,
-    "Serratia": 3,
-    "Saccharopolyspora": 4,
-    "Saccharomyces": 5,
-    "Rothia": 6,
-    "Pseudomonas": 7,
-    "Prevotella": 8,
-    "Photorhabdus": 9,
-    "Oerskovia": 10,
-    "Nocardia": 11,
-    "Morganella": 12,
-    "Moraxella": 13,
-    "Micrococcus": 14,
-    "Methylobacterium": 15,
-    "Massilia": 16,
-    "Leptospira": 17,
-    "Leifsonia": 18,
-    "Kingella": 19,
-    "Helicobacter": 20,
-    "Hafnia": 21,
-    "Gemella": 22,
-    "Frankia": 23,
-    "Finegoldia": 24,
-    "Enterobacter": 25,
-    "Eikenella": 26,
-    "Cryptococcus": 27,
-    "Chromobacterium": 28,
-    "Cellulomonas": 29,
-    "Cardiobacterium": 30,
-    "Candida": 31,
-    "Brucella": 32,
-    "Brevibacterium": 33,
-    "Bordetella": 34,
-    "Bilophila": 35,
-    "Bartonella": 36,
-    "Borrelia": 37,
-    "Bacteroides": 38,
-    "Azotobacter": 39,
-    "Abiotrophia": 40,
-    "Acidaminococcus": 41,
-    "Actinobacillus": 42,
-    "Actinomyces": 43,
-    "Alcaligenes": 44,
-    "Yokenella": 45,
-    "Yersinia": 46,
-    "Veillonella": 47,
-    "Trueperella": 48,
-    "Thermoactinomyces": 49,
-    "Streptomyces": 50,
-    "Streptococcus": 51,
-    "Stenotrophomonas": 52,
-    "Staphylococcus": 53,
-    "Sporolactobacillus": 54,
-    "Spirillum": 55,
-    "Sphingomonas": 56,
-    "Shigella": 57,
-    "Shewanella": 58,
-    "Salmonella": 59,
-    "Ruminococcus": 60,
-    "Rhodococcus": 61,
-    "Rhizobium": 62,
     "Raoultella": 63,
-    "Ralstonia": 64,
-    "Psychrobacter": 65,
-    "Providencia": 66,
-    "Proteus": 67,
-    "Porphyromonas": 68,
-    "Pluralibacter": 69,
-    "Peptostreptococcus": 70,
-    "Peptoniphilus": 71,
-    "Pasteurella": 72,
-    "Parvimonas": 73,
-    "Paenibacillus": 74,
-    "Neisseria": 75,
-    "Mycobacterium": 76,
-    "Myroides": 77,
-    "Lactobacillus": 78,
-    "Leclercia": 79,
-    "Legionella": 80,
-    "Leuconostoc": 81,
-    "Listeria": 82,
-    "Kosakonia": 83,
-    "Kocuria": 84,
-    "Kluyvera": 85,
-    "Klebsiella": 86,
-    "Haemophilus": 87,
-    "Geobacillus": 88,
-    "Gardnerella": 89,
-    "Fusobacterium": 90,
-    "Flavobacterium": 91,
-    "Exiguobacterium": 92,
-    "Eubacterium": 93,
-    "Escherichia": 94,
-    "Erysipelothrix": 95,
-    "Enterococcus": 96,
-    "Elizabethkingia": 97,
-    "Edwardsiella": 98,
-    "Cutibacterium": 99,
-    "Cupriavidus": 100,
-    "Cronobacter": 101,
-    "Corynebacterium": 102,
-    "Comamonas": 103,
-    "Clostridium": 104,
-    "Clostridioides": 105,
-    "Citrobacter": 106,
-    "Cedecea": 107,
-    "Capnocytophaga": 108,
-    "Campylobacter": 109,
-    "Burkholderia": 110,
-    "Bergeyella": 111,
-    "Bacillus": 112,
-    "Arcanobacterium": 113,
-    "Anaerococcus": 114,
-    "Alicyclobacillus": 115,
-    "Aggregatibacter": 116,
-    "Aeromonas": 117,
-    ":": 118,
-    "Aerococcus": 119,
-    "Acinetobacter": 120,
-    "Acidovorax": 121,
-    "Achromobacter": 122
   },
   "idx_to_genus": {
-    "0": "Weissella",
-    "1": "Vibrio",
-    "2": "Spiroplasma",
-    "3": "Serratia",
-    "4": "Saccharopolyspora",
-    "5": "Saccharomyces",
-    "6": "Rothia",
-    "7": "Pseudomonas",
-    "8": "Prevotella",
-    "9": "Photorhabdus",
-    "10": "Oerskovia",
-    "11": "Nocardia",
-    "12": "Morganella",
-    "13": "Moraxella",
-    "14": "Micrococcus",
-    "15": "Methylobacterium",
-    "16": "Massilia",
-    "17": "Leptospira",
-    "18": "Leifsonia",
-    "19": "Kingella",
-    "20": "Helicobacter",
-    "21": "Hafnia",
-    "22": "Gemella",
-    "23": "Frankia",
-    "24": "Finegoldia",
-    "25": "Enterobacter",
-    "26": "Eikenella",
-    "27": "Cryptococcus",
-    "28": "Chromobacterium",
-    "29": "Cellulomonas",
-    "30": "Cardiobacterium",
-    "31": "Candida",
-    "32": "Brucella",
-    "33": "Brevibacterium",
-    "34": "Bordetella",
-    "35": "Bilophila",
-    "36": "Bartonella",
-    "37": "Borrelia",
-    "38": "Bacteroides",
-    "39": "Azotobacter",
-    "40": "Abiotrophia",
-    "41": "Acidaminococcus",
-    "42": "Actinobacillus",
-    "43": "Actinomyces",
-    "44": "Alcaligenes",
-    "45": "Yokenella",
-    "46": "Yersinia",
-    "47": "Veillonella",
-    "48": "Trueperella",
-    "49": "Thermoactinomyces",
-    "50": "Streptomyces",
-    "51": "Streptococcus",
-    "52": "Stenotrophomonas",
-    "53": "Staphylococcus",
-    "54": "Sporolactobacillus",
-    "55": "Spirillum",
-    "56": "Sphingomonas",
-    "57": "Shigella",
-    "58": "Shewanella",
-    "59": "Salmonella",
-    "60": "Ruminococcus",
-    "61": "Rhodococcus",
-    "62": "Rhizobium",
     "63": "Raoultella",
-    "64": "Ralstonia",
-    "65": "Psychrobacter",
-    "66": "Providencia",
-    "67": "Proteus",
-    "68": "Porphyromonas",
-    "69": "Pluralibacter",
-    "70": "Peptostreptococcus",
-    "71": "Peptoniphilus",
-    "72": "Pasteurella",
-    "73": "Parvimonas",
-    "74": "Paenibacillus",
-    "75": "Neisseria",
-    "76": "Mycobacterium",
-    "77": "Myroides",
-    "78": "Lactobacillus",
-    "79": "Leclercia",
-    "80": "Legionella",
-    "81": "Leuconostoc",
-    "82": "Listeria",
-    "83": "Kosakonia",
-    "84": "Kocuria",
-    "85": "Kluyvera",
-    "86": "Klebsiella",
-    "87": "Haemophilus",
-    "88": "Geobacillus",
-    "89": "Gardnerella",
-    "90": "Fusobacterium",
-    "91": "Flavobacterium",
-    "92": "Exiguobacterium",
-    "93": "Eubacterium",
-    "94": "Escherichia",
-    "95": "Erysipelothrix",
-    "96": "Enterococcus",
-    "97": "Elizabethkingia",
-    "98": "Edwardsiella",
-    "99": "Cutibacterium",
-    "100": "Cupriavidus",
-    "101": "Cronobacter",
-    "102": "Corynebacterium",
-    "103": "Comamonas",
-    "104": "Clostridium",
-    "105": "Clostridioides",
-    "106": "Citrobacter",
-    "107": "Cedecea",
-    "108": "Capnocytophaga",
-    "109": "Campylobacter",
-    "110": "Burkholderia",
-    "111": "Bergeyella",
-    "112": "Bacillus",
-    "113": "Arcanobacterium",
-    "114": "Anaerococcus",
-    "115": "Alicyclobacillus",
-    "116": "Aggregatibacter",
-    "117": "Aeromonas",
-    "118": ":",
-    "119": "Aerococcus",
-    "120": "Acinetobacter",
-    "121": "Acidovorax",
-    "122": "Achromobacter"
   },
   "n_features": 73,
-  "num_classes": 123,
   "metrics": {
-    "train_accuracy": 0.9825085324232082,
-    "valid_accuracy": 0.9642248722316865,
-    "best_iteration": 295
   },
   "feature_schema_path": "data/feature_schema.json",
   "feature_names": [

 {
   "genus_to_idx": {
+    "Staphylococcus": 0,
+    "Salmonella": 1,
+    "Listeria": 2,
+    "Enterobacter": 3,
+    "Pseudomonas": 4,
+    "Streptococcus": 5,
+    "Enterococcus": 6,
+    "Bacillus": 7,
+    "Shigella": 8,
+    "Escherichia": 9,
+    "Klebsiella": 10,
+    "Proteus": 11,
+    "Vibrio": 12,
+    "Neisseria": 13,
+    "Campylobacter": 14,
+    "Clostridium": 15,
+    "Corynebacterium": 16,
+    "Legionella": 17,
+    "Mycobacterium": 18,
+    "Bacteroides": 19,
+    "Micrococcus": 20,
+    "Erysipelothrix": 21,
+    "Haemophilus": 22,
+    "Aeromonas": 23,
+    "Yersinia": 24,
+    "Acinetobacter": 25,
+    "Serratia": 26,
+    "Morganella": 27,
+    "Providencia": 28,
+    "Burkholderia": 29,
+    "Helicobacter": 30,
+    "Actinomyces": 31,
+    "Nocardia": 32,
+    "Pasteurella": 33,
+    "Citrobacter": 34,
+    "Leptospira": 35,
+    "Alcaligenes": 36,
+    "Shewanella": 37,
+    "Edwardsiella": 38,
+    "Chromobacterium": 39,
+    "Lactobacillus": 40,
+    "Propionibacterium": 41,
+    "Peptostreptococcus": 42,
+    "Veillonella": 43,
+    "Fusobacterium": 44,
+    "Eubacterium": 45,
+    "Halomonas": 46,
+    "Psychrobacter": 47,
+    "Rhodococcus": 48,
+    "Mycoplasma": 49,
+    "Bordetella": 50,
+    "Stenotrophomonas": 51,
+    "Ralstonia": 52,
+    "Achromobacter": 53,
+    "Brucella": 54,
+    "Arthrobacter": 55,
+    "Flavobacterium": 56,
+    "Oerskovia": 57,
+    "Sphingomonas": 58,
+    "Comamonas": 59,
+    "Thermococcus": 60,
+    "Elizabethkingia": 61,
+    "Hafnia": 62,
     "Raoultella": 63,
+    "Ochrobactrum": 64,
+    "Roseomonas": 65,
+    "Actinobacillus": 66,
+    "Gemella": 67,
+    "Rothia": 68,
+    "Carnobacterium": 69,
+    "Plesiomonas": 70,
+    "Janthinobacterium": 71,
+    "Paenibacillus": 72,
+    "Moraxella": 73,
+    "Aerococcus": 74,
+    "Kocuria": 75,
+    "Leuconostoc": 76,
+    "Arcanobacterium": 77,
+    "Gardnerella": 78,
+    "Porphyromonas": 79,
+    "Prevotella": 80,
+    "Pediococcus": 81,
+    "Weissella": 82,
+    "Lactococcus": 83,
+    "Microbacterium": 84,
+    "Clostridioides": 85,
+    "Cronobacter": 86,
+    "Rhizobium": 87,
+    "Azotobacter": 88,
+    "Spirillum": 89,
+    "Candida": 90,
+    "Cryptococcus": 91,
+    "Saccharomyces": 92,
+    "Rickettsia": 93,
+    "Borrelia": 94,
+    "Chlamydia": 95,
+    "Acidaminococcus": 96,
+    "Bartonella": 97,
+    "Coxiella": 98,
+    "Kingella": 99,
+    "Eikenella": 100,
+    "Bilophila": 101,
+    "Anaerococcus": 102,
+    "Finegoldia": 103,
+    "Parvimonas": 104,
+    "Ruminococcus": 105,
+    "Cutibacterium": 106,
+    "Exiguobacterium": 107,
+    "Kluyvera": 108,
+    "Pluralibacter": 109,
+    "Massilia": 110,
+    "Methylobacterium": 111,
+    "Cupriavidus": 112,
+    "Acidovorax": 113,
+    "Geobacillus": 114,
+    "Trueperella": 115,
+    "Streptomyces": 116,
+    "Thermoactinomyces": 117,
+    "Capnocytophaga": 118,
+    "Cardiobacterium": 119,
+    "Yokenella": 120,
+    "Brevibacterium": 121,
+    "Peptoniphilus": 122,
+    "Weisella": 123,
+    "Saccharopolyspora": 124,
+    "Frankia": 125,
+    "Spiroplasma": 126,
+    "Cedecea": 127,
+    "Photorhabdus": 128,
+    "Abiotrophia": 129,
+    "Cellulomonas": 130,
+    "Leifsonia": 131,
+    "Alicyclobacillus": 132,
+    "Sporolactobacillus": 133,
+    "Leclercia": 134,
+    "Kosakonia": 135,
+    "Bergeyella": 136,
+    "Myroides": 137,
+    "Aggregatibacter": 138,
+    ":": 139
   },
   "idx_to_genus": {
+    "0": "Staphylococcus",
+    "1": "Salmonella",
+    "2": "Listeria",
+    "3": "Enterobacter",
+    "4": "Pseudomonas",
+    "5": "Streptococcus",
+    "6": "Enterococcus",
+    "7": "Bacillus",
+    "8": "Shigella",
+    "9": "Escherichia",
+    "10": "Klebsiella",
+    "11": "Proteus",
+    "12": "Vibrio",
+    "13": "Neisseria",
+    "14": "Campylobacter",
+    "15": "Clostridium",
+    "16": "Corynebacterium",
+    "17": "Legionella",
+    "18": "Mycobacterium",
+    "19": "Bacteroides",
+    "20": "Micrococcus",
+    "21": "Erysipelothrix",
+    "22": "Haemophilus",
+    "23": "Aeromonas",
+    "24": "Yersinia",
+    "25": "Acinetobacter",
+    "26": "Serratia",
+    "27": "Morganella",
+    "28": "Providencia",
+    "29": "Burkholderia",
+    "30": "Helicobacter",
+    "31": "Actinomyces",
+    "32": "Nocardia",
+    "33": "Pasteurella",
+    "34": "Citrobacter",
+    "35": "Leptospira",
+    "36": "Alcaligenes",
+    "37": "Shewanella",
+    "38": "Edwardsiella",
+    "39": "Chromobacterium",
+    "40": "Lactobacillus",
+    "41": "Propionibacterium",
+    "42": "Peptostreptococcus",
+    "43": "Veillonella",
+    "44": "Fusobacterium",
+    "45": "Eubacterium",
+    "46": "Halomonas",
+    "47": "Psychrobacter",
+    "48": "Rhodococcus",
+    "49": "Mycoplasma",
+    "50": "Bordetella",
+    "51": "Stenotrophomonas",
+    "52": "Ralstonia",
+    "53": "Achromobacter",
+    "54": "Brucella",
+    "55": "Arthrobacter",
+    "56": "Flavobacterium",
+    "57": "Oerskovia",
+    "58": "Sphingomonas",
+    "59": "Comamonas",
+    "60": "Thermococcus",
+    "61": "Elizabethkingia",
+    "62": "Hafnia",
     "63": "Raoultella",
+    "64": "Ochrobactrum",
+    "65": "Roseomonas",
+    "66": "Actinobacillus",
+    "67": "Gemella",
+    "68": "Rothia",
+    "69": "Carnobacterium",
+    "70": "Plesiomonas",
+    "71": "Janthinobacterium",
+    "72": "Paenibacillus",
+    "73": "Moraxella",
+    "74": "Aerococcus",
+    "75": "Kocuria",
+    "76": "Leuconostoc",
+    "77": "Arcanobacterium",
+    "78": "Gardnerella",
+    "79": "Porphyromonas",
+    "80": "Prevotella",
+    "81": "Pediococcus",
+    "82": "Weissella",
+    "83": "Lactococcus",
+    "84": "Microbacterium",
+    "85": "Clostridioides",
+    "86": "Cronobacter",
+    "87": "Rhizobium",
+    "88": "Azotobacter",
+    "89": "Spirillum",
+    "90": "Candida",
+    "91": "Cryptococcus",
+    "92": "Saccharomyces",
+    "93": "Rickettsia",
+    "94": "Borrelia",
+    "95": "Chlamydia",
+    "96": "Acidaminococcus",
+    "97": "Bartonella",
+    "98": "Coxiella",
+    "99": "Kingella",
+    "100": "Eikenella",
+    "101": "Bilophila",
+    "102": "Anaerococcus",
+    "103": "Finegoldia",
+    "104": "Parvimonas",
+    "105": "Ruminococcus",
+    "106": "Cutibacterium",
+    "107": "Exiguobacterium",
+    "108": "Kluyvera",
+    "109": "Pluralibacter",
+    "110": "Massilia",
+    "111": "Methylobacterium",
+    "112": "Cupriavidus",
+    "113": "Acidovorax",
+    "114": "Geobacillus",
+    "115": "Trueperella",
+    "116": "Streptomyces",
+    "117": "Thermoactinomyces",
+    "118": "Capnocytophaga",
+    "119": "Cardiobacterium",
+    "120": "Yokenella",
+    "121": "Brevibacterium",
+    "122": "Peptoniphilus",
+    "123": "Weisella",
+    "124": "Saccharopolyspora",
+    "125": "Frankia",
+    "126": "Spiroplasma",
+    "127": "Cedecea",
+    "128": "Photorhabdus",
+    "129": "Abiotrophia",
+    "130": "Cellulomonas",
+    "131": "Leifsonia",
+    "132": "Alicyclobacillus",
+    "133": "Sporolactobacillus",
+    "134": "Leclercia",
+    "135": "Kosakonia",
+    "136": "Bergeyella",
+    "137": "Myroides",
+    "138": "Aggregatibacter",
+    "139": ":"
   },
   "n_features": 73,
+  "num_classes": 140,
   "metrics": {
+    "train_accuracy": 0.9869916267942583,
+    "valid_accuracy": 0.9509569377990431,
+    "best_iteration": 270
   },
   "feature_schema_path": "data/feature_schema.json",
   "feature_names": [