Contrastive losses as generalized models of global epistasis
This work addresses a central task in protein engineering for researchers, offering a more efficient method to estimate fitness functions from limited data, though it appears incremental as it builds on existing global epistasis models.
The paper tackles the problem of inferring multimodal fitness functions from experimental data in protein engineering by showing that minimizing supervised contrastive losses, like the Bradley-Terry loss, effectively extracts sparse latent functions in global epistasis models. The result is consistently improved performance on benchmark tasks, with contrastive losses outperforming Mean Squared Error (MSE) in regimes where MSE is ineffective.
Fitness functions map large combinatorial spaces of biological sequences to properties of interest. Inferring these multimodal functions from experimental data is a central task in modern protein engineering. Global epistasis models are an effective and physically-grounded class of models for estimating fitness functions from observed data. These models assume that a sparse latent function is transformed by a monotonic nonlinearity to emit measurable fitness. Here we demonstrate that minimizing supervised contrastive loss functions, such as the Bradley-Terry loss, is a simple and flexible technique for extracting the sparse latent function implied by global epistasis. We argue by way of a fitness-epistasis uncertainty principle that the nonlinearities in global epistasis models can produce observed fitness functions that do not admit sparse representations, and thus may be inefficient to learn from observations when using a Mean Squared Error (MSE) loss (a common practice). We show that contrastive losses are able to accurately estimate a ranking function from limited data even in regimes where MSE is ineffective and validate the practical utility of this insight by demonstrating that contrastive loss functions result in consistently improved performance on benchmark tasks.