MapPFN: Learning Causal Perturbation Maps in Context

Marvin Sextro, Weronika Kłos, Gabriel Dernbach
1/28/2026
cs.LG

Official code: marvinsxtr/MapPFN (6 stars, MIT, last commit 2026-04-22).

Abstract

Planning effective interventions in biological systems requires treatment-effect models that adapt to unseen biological contexts by identifying their specific underlying mechanisms. Yet single-cell perturbation datasets span only a handful of biological contexts, and existing methods cannot leverage new interventional evidence at inference time to adapt beyond their training data. To meta-learn a perturbation effect estimator, we present MapPFN, a prior-data fitted network (PFN) pretrained on synthetic data generated from a prior over causal perturbations. Given a set of experiments, MapPFN uses in-context learning to predict post-perturbation distributions, without gradient-based optimization. Despite being pretrained on in silico gene knockouts alone, MapPFN identifies differentially expressed genes, matching the performance of models trained on real single-cell data. Our code and data are available at https://github.com/marvinsxtr/MapPFN.

View on arXivView PDF

Code Implementations(1)

marvinsxtr/MapPFNOfficial100%
MIT

MapPFN: Learning Causal Perturbation Maps in Context

62Python, DockerfileJan 27, 20266 months agoMIT

Similar Papers

Do-PFN: In-Context Learning for Causal Effect Estimation

86% match2025

Identifying biological perturbation targets through causal differential networks

84% match2024

Learning to refine domain knowledge for biological network inference

82% match2024

Generative Intervention Models for Causal Perturbation Modeling

82% match2024

The CausalBench challenge: A machine learning contest for gene network inference from single-cell perturbation data

81% match2023

CausalPFN: Amortized Causal Effect Estimation via In-Context Learning

81% match2 repos1212025

Cite this paper

@article{sextro2026mappfn,
  title  = {MapPFN: Learning Causal Perturbation Maps in Context},
  author = {Marvin Sextro and Weronika Kłos and Gabriel Dernbach},
  year   = {2026},
  eprint = {2601.21092},
  archivePrefix = {arXiv},
  url    = {http://arxiv.org/abs/2601.21092v1}
}

Add a badge to your README

Shows whether official code is linked for this paper.

Code of Paper badge
[![Code of Paper](https://codeofpaper.com/badge/2601.21092.svg)](https://codeofpaper.com/paper/2601.21092)

Discussion