Metappuccino: Large Language Model-driven Reconstruction of Sequence Read Archive Metadata for Cancer Research
Preprint
Abstract
Motivation: High-throughput RNA-sequencing has significantly advanced transcriptomic profiling in on-cology. Millions of RNA-seq datasets have accumulated in public databases such as the Sequence Read Archive-SRA. However, fragmented, ambiguous or missing metadata can severely limit accurate cohort selection, introduce bias and delay discoveries. Results: To address these issues, we introduce Metappuccino : a metadata enrichment tool based on a fine-tuned Mistral-7B-Instruct large language model with low-rank-adaptation (LoRA). Metappuccino can extract or infer 19 key metadata classes (e.g. organ, disease, cell type) from unstructured text. Fine-tuning was conducted with careful partitioning and training design to preserve the model’s generalisation capacity, reduce data leakage, and ensure robust, context-aware inference rather than memorisation. When possible, the inferred outputs are mapped to standardised ontologies, such as Cellosaurus, Disease Ontology and Uberon, to produce consistent metadata. As a result, the fine-tuned model achieves significantly improved class prediction accuracy over the base model, performing at least as well as recent large open-source models. Furthermore, it reduces inference time by up to at least two compared to the baseline models. As a pipeline, Metappuccino complements the LLM with well-established Natural Language Processing techniques from the literature to further improve performance. By enriching the metadata of under-annotated sequences, Metappuccino creates greater value from public RNA-seq datasets, with potential applications extending beyond oncology transcriptomics. Availability and Implementation: The source code of Metappuccino is available on GitHub : github. com/chumphati/Metappuccino. The fine-tuned LLM, MetappuccinoLLModel, is available on Hugging Face : huggingface.co/chumphati/MetappuccinoLLModel. Both repositories are released under Apache-2.0 license. Contact: fiona.hak@i2bc.paris-saclay.fr , daniel.gautheret@universite-paris-saclay.fr , melina.gallopin@i2bc.paris-saclay.frKeywords
All Related Versions
- Version 1, 2025-11-24, HAL (Unconfirmed version)
- Published version: Bioinformatics, 42 (5).
This publication has 15 references indexed in Scilit:
- A computational framework for extracting biological insights from SRA cancer dataScientific Reports, 2025
- Extraction of biological terms using large language models enhances the usability of metadata in the BioSample databaseGigaScience, 2025
- Annotating publicly-available samples and studies using interpretable modeling of unstructured metadatabioRxiv, 2024
- ChIP-GPT: a managed large language model for robust data extraction from biomedical database recordsBriefings in Bioinformatics, 2024
- Metadata integrity in bioinformatics: Bridging the gap between data and knowledgeComputational and Structural Biotechnology Journal, 2023
- Increasing metadata coverage of SRA BioSample entries using deep learning–based named entity recognitionDatabase: The Journal of Biological Databases and Curation, 2021
- grabseqs: simple downloading of reads and metadata from multiple next-generation sequencing data repositoriesBioinformatics, 2020
- MetaSRA: normalized human sample-specific metadata for the Sequence Read ArchiveBioinformatics, 2017
- Disease Ontology 2015 update: an expanded and updated database of human diseases for linking biomedical knowledge through disease dataNucleic Acids Research, 2014
- The Sequence Read ArchiveNucleic Acids Research, 2010