Example dataset and end-to-end example

This page demonstrates creating a tiny synthetic dataset and running the main inference flow. Use this as a minimal reproducible example for testing.

import numpy as np
import scanpy as sc
from anndata import AnnData
from Compocyte.core.hierarchical_classifier import HierarchicalClassifier

# Create tiny synthetic data (10 cells, 5 genes)
X = np.random.poisson(1.0, size=(10, 5)).astype(float)
obs = {"cell_type": ["A"]*5 + ["B"]*5}
var = {"gene_ids": [f"g{i}" for i in range(5)]}
adata = AnnData(X=X)
adata.obs["cell_type"] = obs["cell_type"]

# Initialize classifier (no models present; this is only to exercise API)
hc = HierarchicalClassifier(save_path="./tmp_model")
hc.load_adata(adata)

# This example won't train models — it shows usage of API methods
try:
    hc.predict_all_child_nodes('Root')
except Exception as e:
    print('Expected error (no models):', e)

Notes

• For realistic analyses use real single-cell h5ad files and train or load pretrained models before calling predict_all_child_nodes().