API

Interact with HEST-1k

See tutorial 2. Interacting with HEST.

iter_hest(hest_dir[, id_list])

Iterate through HEST samples contained in hest_dir

Run HEST-Benchmark

See tutorial 4. Running HEST Benchmark.

benchmark(encoder, enc_transf, precision[, ...])

Benchmark a patch encoder on HEST-bench

HESTData class

Core object representing a (pooled) Spatial Transcriptomics sample along with a full resolution H&E image and associated metadata. See tutorial 2. Interacting with HEST.

HESTData(adata, img, pixel_size[, meta, ...])

Object representing a (pooled) Spatial Transcriptomics sample along with a full resolution H&E image and associated metadata

Pooling of transcripts, binning

Methods used to pool Xenium transcripts and Visium-HD bins into square bins of custom size

`pool_transcripts_xenium`(df, pixel_size_he[, ...])	Pool a xenium transcript dataframe by square spots of spot_size_um micrometers.
`pool_bins_visiumhd`(adata, pixel_size[, ...])	Pools Visium HD bins from an initial resolution (src_bin_size_um) into larger square spots of spot_size_um.
`pool_bins_visiumhd_per_cell`(nuc_seg, ...[, ...])	Pool Visium-hd bins per cell.

CellViT segmentation

Simplified API for nuclei segmentation

segment_cellvit(wsi_path, name[, ...])

Segment nuclei with CellViT

Gene names manipulation

`unify_gene_names`(adata[, species, drop])	unify gene names by resolving aliases
`ensembl_id_to_gene`(st[, filter_na])	Converts ensemble gene IDs of a HESTData object using Biomart annotations and filter out genes with no matching Ensembl ID

Readers to expand HEST-1k

Readers to expand HEST-1k with additional samples. See tutorial 3. Assembling HEST Data.

`Reader`()	ST/H&E reader
`VisiumReader`()	10x Genomics Visium reader
`XeniumReader`()	10x Xenium reader
`VisiumHDReader`()	10x Genomics Visium-HD reader
`STReader`()	Legacy Spatial Transcriptomics reader

IO

`GDFReader`()	Lazily read shapes such that read_gdf is called at compute time
`XeniumParquetCellReader`([pixel_size_morph, ...])	Xenium parquet shape reader
`GDFParquetCellReader`([use_dask])	Geopandas parquet shape reader
`XeniumTranscriptsReader`(pixel_size_morph[, ...])	Xenium transcript shape reader
`HESTXeniumTranscriptsReader`([use_dask])	HEST Xenium transcript reader
`write_geojson`(gdf, path)	Write a (dask) geodataframe in optimized QuPath geojson detection format.

Batch effect visualization/correction

`filter_hest_stromal_housekeeping`(meta_df, ...)	Filter the genes of HESTData samples, such that: - only stable housekeeping genes are kept (see assets/MostStable_{species}.csv).
`get_silhouette_score`(adata_list, labels[, ...])	Compute the silhouette score for adata_list, cluster memberships are passed in labels (len(labels) == len(adata_list))
`plot_umap`(adata_list, labels, plot_path[, ...])	Create UMAP plot (n=2) for adata_list, cluster memberships are passed in labels (len(labels) == len(adata_list))
`correct_batch_effect`(adata_list[, batch, ...])	Apply a batch effect correction method to a list of Spatial Transcriptomics expressions

Miscellaneous

`tiff_save`(img, save_path, pixel_size[, ...])	Save an image stored in a numpy array to the generic tiff format
`autoalign_visium`(fullres_img[, save_dir, name])	Automatically find the spot alignment based on an image with apparent Visium fiducials
`write_10X_h5`(adata, file)	Writes adata to a 10X-formatted h5 file.
`find_pixel_size_from_spot_coords`(my_df[, ...])	Estimate the pixel size of an image in um/px given a dataframe containing the spot coordinates in that image
`get_k_genes`(adata_list, k, criteria[, ...])	Get the top-k genes according to some criteria in common genes across multiple samples.