LYNX

LYNX is a method for spatially-resolved gradient inference with multi-omic integration. It learns a shared latent representation from paired spatial modalities (e.g. spatial transcriptomics & metabolomics, histology, or protein abundance) using a variational graph auto-encoder on a spatial hetero-graph, then infers continuous spatial gradients, discrete tissue zones, and cell–cell interactions on top of that representation.

Get started

• Overview — what LYNX does and how the pipeline fits together.

• Installation — set up LYNX with conda (preferred) or pip.

• Tutorials — worked examples across three multi-omic datasets.

• API reference — the lynx public API.

Tutorials at a glance

Tutorial	Modalities
Liver — Xenium transcriptomics + DESI metabolomics	Xenium (transcriptomics) + DESI (metabolomics)
Breast — Xenium transcriptomics + H&E histology	Xenium (transcriptomics) + histology
Thymus — spatial RNA + protein (CITE-seq)	Spatial RNA + protein (Stereo-CITE-seq)