OmicsLM: A Multimodal Large Language Model for Multi-Sample Omics Reasoning

Abstract

Large language models can reason over biological knowledge, but they are not naturally equipped to work with quantitative omics measurements. Existing approaches either convert expression profiles into long textual representations, often losing quantitative information, or connect omics encoders to language models for narrow tasks. We introduce OmicsLM, a 9B multimodal language model that represents each bulk or single-cell transcriptome as a compact continuous embedding that can be interleaved with natural-language instructions and with other samples in the same context. An key contribution is a training corpus that connects experimental measurements with biological knowledge and task supervision. We combine templates linking transcriptomic profiles to annotations and experimental outcomes, reasoning examples connecting evidence across sources, and LLM-generated supervision with task-specific quality checks. The instruction-tuning corpus spans 157 biological and general-purpose task groups. Training includes continued pretraining on 20 billion tokens and supervised fine-tuning on 5.21 million conversations. We further introduce GEO-OmicsQA, a publication-disjoint benchmark of 3,000 questions requiring models to interpret and compare expression profiles from real GEO studies. OmicsLM achieves the best results among the evaluated models on three of four held-out annotation tasks and on all six GEO-OmicsQA variants, and the best average performance among compared methods on PerturbQA perturbation-response prediction. Ablations show a clear advantage for continuous omics representations over a text-only gene-list representation. Together, these results show that quantitative transcriptomic measurements and biological language can be integrated within a general conversational model, enabling both predictive tasks and language-guided interpretation of experimental samples.

Experimental reasoning in large language models
J. Kim, P. A. Romero
bioRxiv20 Oct 2026PreprintCC BYFull textSystems & synthetic biologybioRxiv
Evidence Scaling for Zero-Shot Protein Reasoning with Large Language Models
Z. Hao, C. Wang, D. Li +1
bioRxiv27 Sep 2026PreprintCC BY-NC-NDFull textBioinformatics & computational biologybioRxiv
The Platonic brain bridge hypothesis: human brain networks as an architectural prior for multimodal large language models
Pengfei Zhang, Biao Tian, Xiangang Li +1
arXiv14 Sep 2026PreprintCC BY 4.0Full textNeurosciencearXiv
SAIL: Sparse Autoencoders for Interpretable Alignment of Human Vision and Multimodal Large Language Models
T. Gao, L. Zhang, Y. Wang +4
bioRxiv30 Sep 2026PreprintCC BYFull textNeurosciencebioRxiv
pxGPT: a schema-constrained multimodal large language model framework for plant phenotyping
Po-Xing Zheng, Jheng-Yan Ou, Yu-Lin Wu +4
Bioinformatics20 Oct 2026Peer-reviewedCC BY 4.0Full textPlant biologyCrossrefPMC

Cite

BibTeX RIS

BibTeX

@misc{sypetkowski2026omicslm,
  title = {{OmicsLM: A Multimodal Large Language Model for Multi-Sample Omics Reasoning}},
  author = {Maciej Sypetkowski and Łukasz Smoliński and Remigiusz Kinas and Joanna Krawczyk and Bartłomiej Hofman and Aleksandra Możwiłło and Michał Pikusa and Tomasz Jetka and Rafał Powalski},
  howpublished = {arXiv},
  year = {2026},
  month = oct,
  doi = {10.48550/arxiv.2605.06728},
  eprint = {2605.06728},
  archivePrefix = {arXiv},
  url = {https://arxiv.org/abs/2605.06728},
  note = {CC BY-NC-SA 4.0},
}

RIS

TY  - UNPB
TI  - OmicsLM: A Multimodal Large Language Model for Multi-Sample Omics Reasoning
AU  - Maciej Sypetkowski
AU  - Łukasz Smoli{ń}ski
AU  - Remigiusz Kinas
AU  - Joanna Krawczyk
AU  - Bartłomiej Hofman
AU  - Aleksandra Mo{ż}wiłło
AU  - Michał Pikusa
AU  - Tomasz Jetka
AU  - Rafał Powalski
PY  - 2026
DA  - 2026/10/25
PB  - arXiv
DO  - 10.48550/arxiv.2605.06728
UR  - https://arxiv.org/abs/2605.06728
AB  - Large language models can reason over biological knowledge, but they are not naturally equipped to work with quantitative omics measurements. Existing approaches either convert expression profiles into long textual representations, often losing quantitative information, or connect omics encoders to language models for narrow tasks. We introduce OmicsLM, a 9B multimodal language model that represents each bulk or single-cell transcriptome as a compact continuous embedding that can be interleaved with natural-language instructions and with other samples in the same context. An key contribution is a training corpus that connects experimental measurements with biological knowledge and task supervision. We combine templates linking transcriptomic profiles to annotations and experimental outcomes, reasoning examples connecting evidence across sources, and LLM-generated supervision with task-specific quality checks. The instruction-tuning corpus spans 157 biological and general-purpose task groups. Training includes continued pretraining on 20 billion tokens and supervised fine-tuning on 5.21 million conversations. We further introduce GEO-OmicsQA, a publication-disjoint benchmark of 3,000 questions requiring models to interpret and compare expression profiles from real GEO studies. OmicsLM achieves the best results among the evaluated models on three of four held-out annotation tasks and on all six GEO-OmicsQA variants, and the best average performance among compared methods on PerturbQA perturbation-response prediction. Ablations show a clear advantage for continuous omics representations over a text-only gene-list representation. Together, these results show that quantitative transcriptomic measurements and biological language can be integrated within a general conversational model, enabling both predictive tasks and language-guided interpretation of experimental samples.
ER  -

Abstract

Related

Cite