This is the official pretrained model parameters of PertFormer: Integrated bulk-single-cell foundation model enables zero-shot prediction of functional perturbations and cell fate dynamics.

Perturbation Transformer (PertFormer) is an LLM-based foundation model for zero-shot predictions of functional perturbations and cell fate dynamics. PertFormer has 3 billion parameters and was pretrained on massive bulk and single-cell multiomic data. Typically, without the need of task-specific training, PertFormer can realize the predictions of functional regulations, perturbations of genomic elements, genes, and sequences in a zero-shot manner, including the simulation of cell state transitions and disease treatment.

This repository provides official pretrained model checkpoints for the project. These checkpoint files are stored in .pkl format for compatibility with the original codebase.

Hugging Face automatically marks pickle-based files as unsafe because Python pickle deserialization may execute arbitrary code during loading. This is a general warning about the serialization format itself.

The files in this repository are official checkpoints released by the authors. They are intended to contain pretrained model weights only, and no malicious code is intentionally included. We consider these files safe for their intended use with the original project code.

Please load these checkpoints using the original project code and in a trusted environment.

Souce codes of PertFormer is available at https://github.com/alibaba/damo-pertformer.

The manuscript is currently under review. You can cite the preprint version of this paper on bioRxiv: https://doi.org/10.1101/2024.12.19.629561

Fei Gu (gufei.gf@alibaba-inc.com), Damo Academy, Alibaba Group.

Zikun Yang (yangzikun.yzk@alibaba-inc.com), Damo Academy, Alibaba Group.