Search papers, labs, and topics across Lattice.
AssayRouter addresses the challenge of selecting optimal frozen molecular predictors for new assays with limited labels by treating completed assays as pseudo-targets and quantifying predictor utility via reduction in held-out discovery loss. A shared regressor learns to predict this utility from candidate behavior, enabling a frozen ranking to select four sources and fit a convex combiner for new assays. This approach significantly reduces negative log-likelihood compared to Support-CV, demonstrating that candidate-utility correspondence effectively transfers information across predictor families.
Completed assays provide transferable supervision for routing frozen prediction interfaces, outperforming Support-CV by 0.0409 NLL on external regression tasks.
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.
