Work

CellARK v1.0: Frozen knowledge layer and paper reproduction artifacts

Year: 2026

Type: dataset

Abstract: This dataset release accompanies CellARK, a knowledge-grounded agentic framework for single-cell annotation reasoning. It provides a frozen PostgreSQL knowledge layer, its expression-support partition, source-level provenance metadata, and paper-reproduction artifacts. The database separates canonical biological knowledge and rebuildable indexes from benchmark labels, comparator predictions, and private Agent traces. The release is restored and verified with the scripts in the CellARK source repository. It requires PostgreSQL 17 with pgvector and pgcrypto. Raw copyrighted publications and third-party source files are not redistributed.

Source: Zenodo (CERN European Organization for Nuclear Research)

DOI: 10.5281/zenodo.21443595

Authors: Wang, Ziqi, Yi, Guofu

Institutions: Anhui University, Chuzhou University

Language: English

Topic: Single-cell and spatial transcriptomics

Subfield: Molecular Biology

Field: Biochemistry, Genetics and Molecular Biology

Domain: Life Sciences

Open Access status: green

OpenAlex ID: W7169727800

Cites:

Cited by:

Related to: 10