Assayrouter improves molecular predictor selection using past assay data
AssayRouter: Historical Utility Priors for Frozen Molecular Predictor Routing
Artificial Intelligence
Summary
When scientists want to analyze new molecules, they often have many pre-trained prediction tools but don’t know which ones to trust for the new case. The authors developed AssayRouter, a method that uses information from previous completed tests to figure out which predictors are most useful for new molecular assays. This method learns to estimate the usefulness of each predictor based on how well it would reduce prediction errors on new data. Their tests show this approach helps pick better predictor combinations even for unseen types of prediction tools.
What this means in practice
- •For pharmaceutical data scientists: Select the best pre-trained molecular predictors for new assays without retraining or accessing their training data.
- •For chemical informatics engineers: Build small, accurate prediction models by combining a few effective frozen predictors guided by historical assay information.
Authors
Dong Xu, Zhangfan Yang, Jiantao Wu, Shipeng Zhang, Zexuan Zhu, Jiangqiang Li, Jun Zhang, Junkai Ji
Abstract
Laboratories often face a new molecular assay with 16-64 labels and a bank of predictors whose training data and parameters are unavailable. The practical question is which frozen outputs to include in a small local model. AssayRouter treats completed assays as pseudo-targets and labels each candidate by its post-fit utility: the reduction in held-out discovery loss when the candidate is added to the local target predictor. A shared regressor learns to predict this utility from candidate behavior on the support set, without source identity; on a new assay, one frozen ranking selects four sources and separate labels fit a convex combiner. We train only on completed ChEMBL-MT assays and evaluate 24 external regression assays across six frozen interface families. AssayRouter-C lowers strict four-call negative log-likelihood (NLL) by 0.0409 relative to Support-CV@4. Frozen candidate-label permutations confirm that candidate-utility correspondence carries the transferred information, and leave-one-interface-out training shows that the mapping generalizes to unseen predictor families. Completed assays therefore provide transferable supervision for scarce-label routing through frozen prediction interfaces.