Introduction

Summary

perturb-lib is a library of tools for scalable modeling of biological perturbations from experimental data. It provides an implementation of the Large Perturbation Model (LPM) from the paper “In silico biological discovery with Large Perturbation Models”. Furthermore, perturb-lib is a general framework that systematically collects data, embeddings, models, and evaluation tools useful for perturbation model. It also includes perturb-gym which is a suite for easily configurable and scalable training of perturbation models.