SVC-Probe: A Framework for Evaluating Perturbation Generalization in Spatial Foundation-Model Embeddings
For researchers using spatial foundation models in drug discovery, this work provides a diagnostic to stress-test whether embeddings capture transferable perturbation signals rather than just memorizing drug-specific patterns.
The paper introduces SVC-Probe, a framework to evaluate whether spatial foundation-model embeddings from fluorescence microscopy generalize across drug perturbations. It finds that high condition discrimination accuracy (98.6%) does not imply reliable cross-drug prediction, with cosine similarity dropping from 0.944 in-domain to 0.30 under leave-one-drug-out evaluation, revealing that perturbation generalization is a stricter benchmark.
This work examines perturbation generalization in spatial foundation-model embeddings derived from fluorescence microscopy images. Although these models can discriminate drug conditions accurately, it remains unclear whether the learned representations reflect patterns consistent with expected perturbation axes that transfer across drugs. We introduce SVC-Probe, a perturbation-aware framework that combines Subcellular Embedding Atlas Stability, Mondrian Neighborhood Graphs, and a Foundation Model Perturbation Probe to assess embedding stability, neighborhood rewiring, and centroid prediction under drug treatment. Applied to the CM4AI MDA-MB-468 chemical-perturbation atlas comprising 462 antibody labels and SubCell 1536-dimensional embeddings, SVC-Probe demonstrates that 98.6% three-way condition accuracy does not correlate with reliable cross-drug prediction, with cosine similarity diminishing from 0.944 in-domain to 0.30 under leave-one-drug-out evaluation, constituting a two-drug stress test rather than a general benchmark. Null calibration indicates that raw residual-turnover coupling is largely influenced by generic embedding structure, whereas a drug-specific signal emerges under vorinostat and is consistent with chromatin-related reorganization. In contrast, the paclitaxel axis is not robustly reconstructed, likely due to sparse coverage of microtubule-associated proteins. Together, these results introduce and demonstrate a reusable diagnostic framework for stress-testing spatial virtual-cell representations and indicate that perturbation generalization may serve as a stricter and more informative benchmark than baseline condition discrimination.