StandardScaler preprocessing

Clustering n_neighbors

Clustering min_dist=0.0

Clustering n_components 5-10

HDBSCAN algorithm used

Reproducibility seed

Separate 2D visualization UMAP

Cluster statistics reported

Clustering evaluation

Cosine distance metric

StandardScaler preprocessing

Reproducibility seed

2D visualization output

Fit then transform for queries

Corpus and queries separated before fit

Embedding output saved

Visualization produced

min_dist for visualization

Query neighbor search implemented

Unlabeled marked as -1

Labels passed to UMAP fit

Unlabeled samples in UMAP fit

StandardScaler preprocessing

n_components for feature engineering

Downstream classifier trained

transform() on test data

Labeled-only classifier training

Reproducibility seed

sklearn Pipeline used

StandardScaler in pipeline

UMAP in pipeline

Labels passed to pipeline fit

target_weight set explicitly

n_components for feature engineering

Reproducibility seed

Train/test split

Pipeline predict on test set

Accuracy reported

target_metric set for classification

AlignedUMAP import

List of datasets passed to fit

embeddings_ attribute accessed

StandardScaler preprocessing

alignment_regularisation set

Reproducibility seed

Visualization per timepoint

Multiple timepoint datasets

Embeddings output saved

n_components=2 for visualization

densmap=True enabled

output_dens=True set

rad_orig_ accessed

rad_emb_ accessed

inverse_transform called

StandardScaler preprocessing

Reproducibility seed

Density comparison output

dens_lambda set explicitly

Reconstruction error reported