LAIOR Benchmarks
Review the LAIOR benchmark across 23 single-cell analysis models, 66 datasets, and 24 evaluation metrics.
23
Models
48
RNA
18
ATAC
24
Metrics
Three places to start
Use the homepage for project context, SCPortal for the wider project map, and this site for the benchmark comparisons.
Start with the benchmark question
Jump straight to models, datasets, or metrics depending on what you need to compare.
Compare model families
Review predictive, generative, trajectory, ATAC-specific, geometric, and disentanglement models side by side.
View 23 modelsAudit dataset coverage
Check scRNA-seq and scATAC-seq coverage before choosing an evaluation setup.
Browse 66 datasetsSelect evaluation metrics
Match clustering, embedding, latent-space, and runtime metrics to your comparison.
Review 24 metricsModel Categories
Browse 6 benchmarked model groups, from contrastive learning to disentangled representations.
Predictive Models
Encoder-only models without decoder reconstruction (contrastive learning)
Generative Models
VAE/autoencoder models with encoder + decoder reconstruction (general VAE class)
Trajectory Inference Models
VAE models specialized for developmental trajectory and pseudo-time inference
scATAC-Specific Models
VAE models specifically designed for chromatin accessibility data
Gaussian Geometric Models
VAE models using geometric structures (hyperbolic, mixture models, learnable curvature)
Disentanglement Models
VAE models focused on learning disentangled factor representations
Benchmark Datasets
Curated single-cell datasets used in the LAIOR benchmark.
Evaluation Metrics
24 metrics grouped into 4 evaluation categories.
Clustering & Cell Type Discovery
Supervised metrics comparing predicted clusters to ground truth labels
Embedding Quality (UMAP & t-SNE)
Visualization quality via coranking analysis (4 metrics × 2 methods)
Intrinsic Latent Space (LSE)
Unsupervised geometric, spectral, and topological properties
Computational Efficiency
Training and inference performance
What you can check
Use these views to inspect the benchmark setup and the reported comparisons.
How the models differ
See how LAIOR compares with LiVAE, iVAE, and other baselines, with short notes on each design.
Which datasets are included
Review 66 curated datasets, including scRNA-seq (48) and scATAC-seq (18) collections.
Which results you can check
24 metrics for clustering (6), embedding quality (8), latent space (8), and runtime (2).
Open the model comparison first
The models view is the quickest entry into the LAIOR benchmark, with datasets and metrics one step away.