# Prototype Guided Post-pretraining for Single-Cell Representation Learning

Preprint · Sachini Weerasekara, Natasha Darras, Sagar Kamarthi, Colles Price, Jacqueline Isaacs

- Archive id: arxiv:2605.07938
- Published: 2026-05-08
- arXiv: https://arxiv.org/abs/2605.07938
- Citations: 0
- Source URL: https://arxiv.org/abs/2605.07938
- PDF: https://arxiv.org/pdf/2605.07938
- Provenance: arxiv

## Abstract

Single-cell representation learning (SCRL) from gene expression data offers a way to uncover the complex regulatory logic underlying cellular function. Inspired by large language models in natural language modeling, several single-cell pretrained models have recently been proposed that treat genes as tokens and cells as sentences. However, these models are fundamentally limited by the long-tailed nature of cell-type distributions and struggle to generalize under covariate shifts in gene expression data. While fine-tuning is often used to mitigate these issues, we observe that performance remains bounded. To address this challenge, we introduce CellRefine, a post-pretraining method that operates between the pretraining and fine-tuning stages of a single-cell foundation model. CellRefine uses a multi-faceted objective that incorporates marker-gene sets as structural priors to guide post-pretraining and refine the latent embedding manifold of cells. Across multiple computational biology tasks, empirical results show that CellRefine consistently improves downstream performance, yielding gains up to 15%.

_Abstract reproduced under CC0-1.0 (arXiv descriptive metadata)._

Topics: [Single-Cell & Spatial Omics](https://space-kitty.com/topics/single-cell-spatial)

---
Space Kitty Archive · https://space-kitty.com · API: https://space-kitty.com/api/v1 · MCP: https://space-kitty.com/mcp · Premium: $0.20/request via x402 (Base Sepolia (testnet)). Metadata: arXiv (CC0), OpenAlex (CC0), Europe PMC (CC BY/CC0 works), ClinicalTrials.gov.