DBCLS BioHackathon 2026 report: RDF publication of Marchantia polymorpha gene annotations, and agent-assisted preparation of DDBJ submissions
Abstract
This report describes work carried out during the DBCLS BioHackathon 2026 (BH26) in Matsuyama, Japan, on three loosely related but complementary threads, all concerned with moving genome annotation data across the boundary between the systems that produce it and the systems that consume it. First, gene annotations of the liverwort *Marchantia polymorpha*, maintained in MarpolBase [@tanizawa2026marpolbase], were converted to RDF: an RDF structure was designed, a converter from GFF3 was written, and the resulting graph was loaded into Virtuoso and deployed to a production endpoint. Second, a prototype agent for preparing submissions to the DNA Data Bank of Japan (DDBJ), provisionally called *DFAST-Agent*, was assembled from a Model Context Protocol (MCP) server [@mcp2026] and a set of skills; submission files were successfully produced through an interactive terminal session, including operations on files held in Kura, DDBJ's authenticated object storage. Third, *Markit*, a tool for registering phylogenetic marker sequences, gained a web interface that was deployed to a staging environment, alongside fixes to its command-line counterpart. Taken together, these threads probe the same question from two sides: how to make annotation data machine-actionable once it has been published, and how to lower the cost of publishing it in the first place.
Related
- Prime-Editing in Marchantia paleacea : Expanding the Genome-Editing Toolbox in Bryophytes
- Expanding the Human Ancestry Ontology to include under-represented populations and ethnicities for broader utility in annotations
- DR-GEM enables self-supervised machine learning for single-cell embeddings and annotations
- An Annotation Consistency Checker for Evidence-Based Evaluation of Untargeted LC-MS/MS Metabolite Annotations
- SureChEMBL in 2026: an open database of chemical and biomedical annotations from the patent literature
BibTeX
@misc{tanizawa2026dbcls,
title = {{DBCLS BioHackathon 2026 report: RDF publication of Marchantia polymorpha gene annotations, and agent-assisted preparation of DDBJ submissions}},
author = {Yasuhiro Tanizawa and Takatomo Fujisawa},
howpublished = {BioHackrXiv},
year = {2026},
month = sep,
doi = {10.37044/osf.io/xev4p_v1},
url = {https://osf.io/preprints/biohackrxiv/xev4p_v1/},
note = {CC BY 4.0},
}
RIS
TY - UNPB
TI - DBCLS BioHackathon 2026 report: RDF publication of Marchantia polymorpha gene annotations, and agent-assisted preparation of DDBJ submissions
AU - Yasuhiro Tanizawa
AU - Takatomo Fujisawa
PY - 2026
DA - 2026/09/25
PB - BioHackrXiv
DO - 10.37044/osf.io/xev4p_v1
UR - https://osf.io/preprints/biohackrxiv/xev4p_v1/
AB - This report describes work carried out during the DBCLS BioHackathon 2026 (BH26) in Matsuyama, Japan, on three loosely related but complementary threads, all concerned with moving genome annotation data across the boundary between the systems that produce it and the systems that consume it. First, gene annotations of the liverwort *Marchantia polymorpha*, maintained in MarpolBase [@tanizawa2026marpolbase], were converted to RDF: an RDF structure was designed, a converter from GFF3 was written, and the resulting graph was loaded into Virtuoso and deployed to a production endpoint. Second, a prototype agent for preparing submissions to the DNA Data Bank of Japan (DDBJ), provisionally called *DFAST-Agent*, was assembled from a Model Context Protocol (MCP) server [@mcp2026] and a set of skills; submission files were successfully produced through an interactive terminal session, including operations on files held in Kura, DDBJ's authenticated object storage. Third, *Markit*, a tool for registering phylogenetic marker sequences, gained a web interface that was deployed to a staging environment, alongside fixes to its command-line counterpart. Taken together, these threads probe the same question from two sides: how to make annotation data machine-actionable once it has been published, and how to lower the cost of publishing it in the first place.
ER -