Python implementation can be found in the 'SIRV' folder. The SIRV function takes as input 1) spatial transcriptomics data as scanpy/scvelo Anndata object, 2) scRNA-seq data as scanpy/scvelo Anndata object, having spliced and unspliced expressions and (optionally) cell label/metadata annotations, 3) number of principal vectors (PVs) needed for integration, 4) column names (identifiers) of the label/metadata annotations to be transferred from scRNA-seq to spatial data (optional).
For full description, please check the SIRV function description in main.py.
The Notebooks folder contains scripts showing how to reproduce the results when testing SIRV on the developing chicken heart, developing mouse brain, and mouse organogenesis datasets.
All processed spatial and scRNA-seq datasets can be downloaded as scanpy Anndata .h5 files from Zenodo
For citation and further information please refer to: "SIRV: Spatial inference of RNA velocity at the single-cell resolution", NAR Genomics and Bioinformatics