data/umap_labels.json

{
  "species": [
    "human",
    "macaque",
    "mouse",
    "rat",
    "dog",
    "cow",
    "pig",
    "chicken",
    "frog",
    "zebrafish",
    "fly",
    "worm",
    "arabidopsis",
    "soybean",
    "tomato",
    "maize",
    "rice",
    "yeast",
    "fission_yeast",
    "candida",
    "aspergillus",
    "neurospora",
    "ecoli",
    "bsubtilis",
    "saureus",
    "sarscov2",
    "hiv1"
  ],
  "biotypes": [
    "protein_coding",
    "lncRNA",
    "snRNA",
    "misc_RNA"
  ],
  "strands": [
    "+",
    "-"
  ],
  "species_kingdom": {
    "human": "vertebrates",
    "macaque": "vertebrates",
    "mouse": "vertebrates",
    "rat": "vertebrates",
    "dog": "vertebrates",
    "cow": "vertebrates",
    "pig": "vertebrates",
    "chicken": "vertebrates",
    "frog": "vertebrates",
    "zebrafish": "vertebrates",
    "fly": "invertebrates",
    "worm": "invertebrates",
    "arabidopsis": "plants",
    "soybean": "plants",
    "tomato": "plants",
    "maize": "plants",
    "rice": "plants",
    "yeast": "fungi",
    "fission_yeast": "fungi",
    "candida": "fungi",
    "aspergillus": "fungi",
    "neurospora": "fungi",
    "ecoli": "bacteria",
    "bsubtilis": "bacteria",
    "saureus": "bacteria",
    "sarscov2": "viruses",
    "hiv1": "viruses"
  },
  "kingdoms": [
    "vertebrates",
    "invertebrates",
    "plants",
    "fungi",
    "bacteria",
    "viruses"
  ],
  "bounds_2d": [
    -13.2362,
    21.4102,
    -12.5187,
    23.0468
  ],
  "bounds_3d": [
    -5.392,
    18.9592,
    -2.8041,
    24.703,
    -11.3491,
    15.3082
  ],
  "length_log10_range": [
    0.7781512503836436,
    6.47142359330933
  ],
  "length_bp_range": [
    6,
    2960899
  ],
  "n_points": 571810,
  "has_3d": false,
  "has_gc": true,
  "has_length": true,
  "has_names": true,
  "source": "HuggingFaceBio/carbon_embeddings (carbon_dana_embeddings_middle)"
}