zmap-tools

zmap-tools is the Python API for the Zebrafish Multi-Atlas Project (ZMAP) — a curated single-cell RNA-seq reference atlas for zebrafish development.

It provides two core workflows:

1. Explore the reference atlas — load the ZMAP reference and browse curated markers or generate publication-ready dotplots.

import zmap

# Fetch data
adata_ref = zmap.ref.load_zmap_h5ad()

# Browse curated marker genes
markers = zmap.ref.load_consensus_markers(groups=['diencephalon'], format='table', n_per_group=10)

# Visualize expression across cell types
zmap.dotplot.gene_view(adata_ref, "sox2")

2. Annotate a query dataset — transfer ZMAP cell-type labels onto your own single-cell data using kNN-based projection.

zmap.predict.annotate_with_zmap(
    adata_query,
    query_raw_counts_source="counts",
    cluster_col="leiden",
)

Key Features

  • Reference loading — download and cache ZMAP H5AD files with persistent Google Drive integration for Colab workflows.

  • Consensus markers — access curated marker gene tables ranked by specificity, contrast, consensus, and prevalence.

  • Dotplots — publication-ready dotplots showing expression across cell types, timepoints, and studies.

  • Label transfer — kNN-based cell-type annotation batch-corected embeddings, with per-cell confidence scoring and QC filters.

Indices and tables