A confound-diagnostic toolkit for in silico perturbation with single-cell foundation models

Deleting a gene token from a cell's input sequence offers a convenient native strategy for in silico perturbation, but the resulting embedding delta may not represent a biological knockout response. Apparent effects can instead reflect gene identity, universal responsiveness, limited tokenization coverage, library-size contamination, or circular state scoring. Here, we present a confound-diagnostic framework combining held-out increment testing, responsiveness adjustment, coverage gating, library-size diagnostics, and de-circularized state-shift analysis, together with a numerically matched reimplementation of frozen Geneformer's perturbation engine. Across Frangieh and Replogle datasets and linear and nonlinear readouts, the native embedding delta provided no reproducible held-out improvement beyond gene identity. Signal-injection calibration showed that the test detected injected residual signal, whereas native increments remained below its detection floor. Matched controls traced apparent positives to raw-count library-size structure, broad responsiveness, and self-referential scoring, while coverage constrained perturbation applicability and estimate stability without establishing biological specificity. This model-adaptable framework helps determine when foundation-model perturbation readouts warrant biological interpretation.