Computational Models

Three computational models are available for analyzing your single-cell RNA-seq data.

Available Models

PCA (Principal Component Analysis)

• Linear method: Captures major sources of variation
• Batch correction: Harmony integration available
• Speed: Fastest processing time
• Memory: Lowest resource requirements
• Interpretability: Component loadings show gene contributions

scVI (Single-cell Variational Inference)

• Generative model: Probabilistic framework for single-cell data
• Census-trained: Uses pre-trained models from reference data
• Classification: Supports cell type prediction
• Organisms: Human and mouse only

TranscriptFormer

• Transformer architecture: Attention-based gene interaction modeling
• Model variants: Sapiens (human), Exemplar (cross-species), Metazoa (broad)
• Classification: Supports cell type prediction
• Resources: Benefits from GPU acceleration