Fig. 2: Performance in predicting post-perturbation gene expression.
From: In silico biological discovery with large perturbation models
The performance of LPM was compared against state-of-the-art baselines across a variety of experimental settings, contexts and for different perturbation types. a, A comparison of methods for post-perturbation expression prediction using z-normalized data including all readouts comparing Pearson correlation (y axis) on held-out test data from eight experimental contexts (x axis) including single-cell (Replogle et al.9), bulk (LINCS7), genetic (CRISPRi and CRISPR-KO) and chemical compound interventions. b,c, In addition, we performed a comparison methods for post-perturbation expression prediction that replicates the preprocessing methodology from Roohani et al.15 and Cui et al.32. In this comparison, we calculated the Pearson correlation between true and predicted changes in log-normalized expression (control versus perturbed) measured on held-out test data for all genes (b) and on the subset of the top 20 differentially expressed transcripts (c) (y axis). Norman et al.76 include both single and multiperturbation data. Embedding (‘emb’ in parentheses) next to a baseline indicates that we used embeddings that were fine-tuned using Catboost. For baselines without this indication, we used author instructions for generating the post-perturbation expression predictions. Not all methods are suitable for all settings that LPM operates on and are therefore not included in all comparisons. Asterisks indicate statistical significance (one-sided Mann–Whitney, *P ≤ 0.05). Dots on top of bars represent random seeds.
