Intel / article

BaseCamp --- An Agentic AI Framework for Automating DNA Sequencing Data Pipelines

Source: arxiv.org
Author: Eranga Bandara, Xueping Liang, Asanga Gunaratna, Tharaka Hewa, Abdul Rahman, Peter Foytik, Safdar H. Bouk, Sachini Rajapakse, Isurunima Kularathna, Pramoda Karunarathna, Chalani Rajapakse, Ng Wee Keong, Kasun De Zoysa, Amin Hass, Wathsala Herath, Ross Gore, Ravi Mukkamala, Nihal Siriwardanagea, Gihan Siriwardanagea, Aruna Withanage, Nilaan Loganathan, Sachin Shetty
Date: Published Sep 25, 2026

Context

BaseCamp proposes making the decisions around DNA-sequencing analysis inspectable. Established analysis tools still perform sequence and identify genetic variants. AI agents choose settings, interpret outputs and refer uncertain decisions to people.

One concrete design requirement is a filtering ledger: it records each threshold, how many variant calls it removed and the rationale for applying it. Calls near a threshold boundary are flagged for human adjudication instead of being silently resolved. This would let a researcher inspect exclusions alongside the surviving results. Several of the paper's evaluation tables and figures are labeled as placeholders, so those numbers do not establish the design's measured performance.

Key takeaways · AI-distilled

BaseCamp's agents do not analyze sequences. Established tools still run alignment, variant calling and annotation; the agents select and configure those tools, interpret their output and decide what follows.
Six agents cover intake and quality control, alignment, variant calling, annotation, cross-stage monitoring and reporting, coordinated by a central reasoning over fine-tuned domain-specialized models.
The system runs locally under orchestration, so no sequencing data leaves the operating environment.
Reported results: -generated configurations agree with expert practice, an explicit filtering ledger records what filtering removes, and cross-stage anomaly detection surfaces conditions that execution monitoring misses. The abstract gives no figures.

Terms in this piece · Glossary

alignment — The work of making AI systems actually pursue what their builders and users intend, rather than something subtly or dangerously different.
LLM — A large language model — the neural network behind tools like Claude and ChatGPT, trained on huge amounts of text to predict what comes next.
human-in-the-loop — Requiring a person's approval at specific points in an automated process, chosen so the irreversible steps are the ones a human sees.
AI agent — An AI system that doesn't just answer once but works toward a goal in a loop — taking actions, reading the results, and deciding what to do next.

Read the source arxiv.org