Source: https://docs.nygen.io/CyteArc/seurat/
Revision: fccee7b6e0c167e47ce605c8d292fba5b567ebfe
Structured page: https://docs.nygen.io/CyteArc/seurat.json

CyteArc for Seurat users

This guide maps familiar Seurat stages to CyteArc’s store-backed API. Use the workflow map below to find the equivalent task, or try the Quick start <https://docs.nygen.io/CyteArc/quickstart> to inspect a prepared result. For optional background on how CyteArc stores and reuses results, see How CyteArc compares with Scanpy and Seurat <https://docs.nygen.io/CyteArc/scanpy-and-seurat>.

Workflow map

The methods below are approximate counterparts rather than identical implementations. After importing an .rds file, the CyteArc column describes analysis on the resulting Zarr store.

Goal	Seurat	CyteArc

Hold the analysis	SeuratObject	DataStore

Load a saved project	readRDS()	inspect_seurat, SeuratReader, SeuratToZarr, then DataStore

Work with modalities	Assays such as RNA and ADT	Assays in the same Zarr store

Select and normalize features	NormalizeData, FindVariableFeatures	ds.features.hvgs, then ds.features.normalize, using exact refs

Scale, reduce, and find neighbours	ScaleData, RunPCA, FindNeighbors	ds.reduction.pca standardizes features by default, followed by CyteArc’s neighbour-graph methods

Embed and cluster	RunUMAP, FindClusters	ds.embeddings.umap, ds.clusters.leiden

Correct batches with Harmony	Harmony integration after PCA	ds.reduction.harmony after PCA, followed by graph construction

Integrate modalities with WNN	FindMultiModalNeighbors	Build neighbours per assay, then call ds.integration.modalities(...); WNN is the default

CyteArc WNN follows the published weighting equations but uses the supplied neighbour candidates rather than Seurat’s wider default search. See Compare multimodal integration methods <https://docs.nygen.io/CyteArc/tutorials/multimodal-diagnostics> before interpreting differences between integration methods. SNN remains available with method="snn".

Import a saved Seurat object

CyteArc imports a saved Seurat object from an .rds file. It reads the on-disk RDS document and does not attach to a live R session. It does not read .h5seurat.

Inspect the RDS file, select importable assays and reductions, then write a Zarr store:

import cytearc

inspection = cytearc.inspect_seurat("pbmc.rds")
with cytearc.SeuratReader(
    "pbmc.rds",
    assays=["RNA"],
    reductions=["pca"],
) as reader:
    imported = cytearc.SeuratToZarr(reader, zarr_loc="pbmc.zarr").dump()
ds = cytearc.DataStore("pbmc.zarr")
imported.activeIdentity, imported.reductionArtifacts["pca"]

The importer brings across supported count layers and literal cell metadata. It returns exact artifact refs for active.ident and selected reductions. Neighbour graphs, images, commands, and most tool slots stay behind. Graphs, clusterings, marker searches, and integrated analyses are rebuilt in CyteArc rather than imported from the RDS object. CyteArc does not write .rds or .h5seurat.

Typical next steps are ds.pipeline.run(), Identify PBMC populations with scRNA-seq <https://docs.nygen.io/CyteArc/tutorials/scrna-seq>, or Build a neighborhood graph stage by stage <https://docs.nygen.io/CyteArc/tutorials/graph-construction>. For multimodal data, build a neighbour artifact per assay before calling integration.modalities. When you only need raw matrices, original 10x HDF5 or Matrix Market counts are still preferable to an RDS export.

To return to Seurat, write H5AD or Matrix Market from CyteArc and convert or import it with the tools used by your R workflow. See Import and export <https://docs.nygen.io/CyteArc/tutorials/import-and-export> for the full Seurat import contract and the other format paths.