Modeling cellular phenotypes is a fundamental challenge in computational systems biology. Accurately predicting cell fate can advance our understanding of both healthy and diseased states and facilitate the identification of novel therapeutic targets. Post-perturbation transcriptomics data are particularly suited for training computational models because the causal relationship between known perturbations and the measured post-perturbation gene expression allows for the modelling of mechanistic processes. However, acquiring such perturbation data is more complex than obtaining baseline (non-perturbed) transcriptomics data.