Prompt-to-Paper: Agentic AI System for Bioinformatics
06:00 · July 8, 2026 · arXiv cs.AI RSS
While recent advances in large language models have enabled end-to-end automated manuscript generation, existing systems suffer from three critical deficiencies: (i) generated claims are not deterministically grounded in verifiable literature, (ii) experimental results are frequently fabricated rather than executed, and (iii) there exists no standardized, multi-dimensional framework to assess whether AI-generated manuscripts meet the quality and rigor required for real-world publication. We present Prompt-to-Paper, a multi-agent framework that directly addresses this evaluation gap through three integrated innovations. First, a deterministic retrieval-augmented generation pipeline with section-aware relevance scoring and snowball citation expansion grounds every claim in a verifiable corpus of 60--100 papers. Second, an autonomous coding agent executes real computational biology experiments replacing synthetic outputs with genuine numerical results. Third, an eight-dimensional automated quality scorer, benchmarked with approximate reference statistics from published papers and augmented with explicit hallucination penalties, provides standardized, reproducible quality assessments. The quality-driven improvement loop uses a context-rich reviser that routes each iteration to one of three researcher actions and fires a deep research cycle every ten iterations to re-run experiments and re-manuscript from stronger outputs. We validate the system on five bioinformatics case studies; all five cases compiled submission-formatted PDFs with zero out-of-range citations. The improvement loop raises manuscript quality by an average of +17.96 points on a 0--100 scale (maximum +26.04. As partial external checks, a human reviewer scored the five manuscripts at an average of 7.0 out of 10. Complete manuscripts are produced at approximately 0.31 USD per paper.
Summary
Prompt-to-Paper is a multi-agent framework that automates the production of submission-ready manuscripts in bioinformatics while addressing three persistent weaknesses of current LLM-based systems: ungrounded claims, fabricated experimental results, and the absence of reproducible quality metrics. The pipeline begins with a deterministic retrieval-augmented generation stage that pulls 60–100 papers from Semantic Scholar, scores each document section-wise against the query, and expands the set through snowball sampling of references and citations. Every generated statement is thereby tied to an explicit, verifiable source rather than model-internal knowledge.
A second agent executes actual computational biology code, inserting measured numerical outcomes into the manuscript instead of synthetic placeholders. An eight-dimensional quality scorer, calibrated against statistics from published papers and equipped with explicit penalties for hallucination, then evaluates the draft. When scores fall short, a context-rich reviser routes the next iteration to one of three targeted actions—adding statistical analysis, retrieving additional literature evidence, or rewriting for clarity—and triggers a deeper research cycle every ten iterations to refresh experiments and regenerate stronger sections.
Validation on five bioinformatics case studies produced complete, formatted PDFs with zero out-of-range citations. The iterative loop raised average manuscript quality by 17.96 points on a 0–100 scale, reaching a maximum gain of 26.04 points. Independent large-language-model rankings aligned with these scores, and a human reviewer assigned an average of 7.0 out of 10. The entire process runs at roughly 0.31 USD per paper.
Why it matters
This research is highly relevant for Dutch AI researchers and bioinformatics practitioners as it introduces a transparent, verifiable approach to AI-assisted research generation. Its focus on eliminating hallucinations and executing real experiments aligns strongly with the Netherlands' emphasis on ethical, trustworthy AI and its robust life sciences sector.
