mlsanityData to 14 Oct 2026

PerturbPFN: Probing the Limits of Synthetic Priors in Drug Perturbation Modelling

arXiv 2607.23447 ↗First posted 26 Jul 2026cs.LGcited by 2

Not on the map. Why?

Abstract (arXiv)

Predicting cellular responses to unseen chemical perturbations is challenging due to unknown targets and mechanisms, high-dimensional expression responses, and limited experimental coverage of the large small-molecule design space. We propose PerturbPFN, a PFN-style amortized model for unknown-target perturbation prediction under a hierarchical synthetic structural prior. Instead of directly regressing high-dimensional expression responses, PerturbPFN infers a latent system graph, sparse atomic intervention targets, and intervention strengths, then propagates their effects through an SCM decoder. The model is trained entirely on prior-predictive synthetic episodes generated from biologically motivated graph and expression simulators, enabling structured in-context learning without test-time gradient updates. We evaluate PerturbPFN on both real single-cell perturbation data and synthetic benchmarks, covering effect prediction, target identification, and regulatory structure discovery. Our results show that PerturbPFN offers a complementary trade-off to specialized baselines, achieving competitive perturbation prediction with low inference cost while exposing interpretable intermediate estimates of targets, strengths, and system structure.

Key facts. Cited by 2 papers in the map (data through 14 Oct 2026). Builds on 23 of its 50 references that are in the map. First posted on arXiv 26 Jul 2026.

Builds on

23 of 50 references.

References · 23

  1. 2018DAGs with NO TEARS: Continuous Optimization for Structure Learningcited by 752
  2. 2017Meta-learners for Estimating Heterogeneous Treatment Effects using Machine Learningcited by 478
  3. 2021Transformers Can Do Bayesian Inferencecited by 360
  4. 2025TabICL: A Tabular Foundation Model for In-Context Learning on Large Datacited by 289
  5. 2026TabICLv2: A better, faster, scalable, and open tabular foundation modelcited by 144
  6. 2025TabPFN-2.5: Advancing the State of the Art in Tabular Foundation Modelscited by 132
  7. 2017Learning Independent Causal Mechanismscited by 105
  8. 2019Permutation-Based Causal Structure Learning with Unknown Intervention Targetscited by 69
  9. 2022Supervised Training of Conditional Monge Mapscited by 61
  10. 2025Do-PFN: In-Context Learning for Causal Effect Estimationcited by 51
  11. 2025CausalPFN: Amortized Causal Effect Estimation via In-Context Learningcited by 49
  12. 2022Predicting Cellular Responses to Novel Drug Perturbations at a Single-Cell Resolutioncited by 45
11 more
  1. 2022Causal de Finetti: On the Identification of Invariant Causal Structure in Exchangeable Datacited by 42
  2. 2023Modelling Cellular Perturbations with the Sparse Additive Mechanism Shift Variational Autoencodercited by 38
  3. 2025Foundation Models for Causal Inference via Prior-Data Fitted Networkscited by 32
  4. 2022BaCaDI: Bayesian Causal Discovery with Unknown Interventionscited by 26
  5. 2023Causal machine learning for single-cell genomicscited by 23
  6. 2024A Meta-Learning Approach to Bayesian Causal Discoverycited by 22
  7. 2025Position: The Future of Bayesian Prediction Is Prior-Fittedcited by 20
  8. 2025Estimating Interventional Distributions with Uncertain Causal Graphs through Meta-Learningcited by 13
  9. 2024Generative Intervention Models for Causal Perturbation Modelingcited by 4
  10. 2025In-silico biological discovery with large perturbation modelscited by 4
  11. 2026MapPFN: Learning Causal Perturbation Maps in Contextcited by 3

Built on by

Cited by year (2026 to 14 Oct)

Citing papers per year: 2026: 2.

2 papers.

  1. 2026Causal Foundation Modelscited by 1
  2. 2026CIDER-FM: Foundation Models for Causal Inference from Diverse Experimental Regimescited by 0