bioRxiv Science⌕ Search

Biology subjects

Takemaru, L.

Publications and source records attributed to Takemaru, L..

2 recordsLinked to original sources
bioRxiv2026

Limits of marker information transfer]{Limits of marker information transfer in cross-panel cytometry imputation

Marker imputation can bridge cytometry datasets acquired with different panels and platforms. Its validity depends on whether common markers retain information about unmeasured target markers and whether those relationships transfer across assay contexts. We evaluated these limits in two paired flow cytometry-CyTOF cohorts comprising 134 patients and 257 specimens. A patient-held-out framework distinguished within-platform same-cell reconstruction from cross-platform distribution translation without cell correspondence. We compared conventional machine-learning and deep generative estimators with median imputation and used permutation controls to estimate the contributions of patient matching and source-exclusive markers. Evaluation covered marker values, specimen distributions, phenotype utility, and directional predictive relationships among markers. Within platforms, exact k-nearest neighbors reduced mean absolute error by 12.8-35.0% relative to median imputation. However, the median-based reconstructions were contracted, with a median ratio of predicted to observed interquartile range below 1 in every context and evaluable population. Patient matching improved cross-platform distribution recovery, whereas adding source-exclusive markers or higher-capacity estimators did not consistently produce net gains. In the independently annotated acute myeloid leukemia (AML) cohort, reconstructed markers largely preserved broad-lineage classification but did not recover fine-phenotype information available through direct measurement. In AML-CyTOF, the measured full panel outperformed the reconstructed representation by 10.25 percentage points for fine phenotypes. Predictive relationships among markers were context-specific. Most targets were better explained by several markers than by one. Marker selection based on these relationships showed a trade-off between panel size and pairwise coverage. Paired samples and imputation can therefore connect selected marker distributions across panels but should not be treated as substitutes for direct measurement. Cross-panel cytometry should be evaluated with patient-held-out tests matched to the resolution of available ground truth and the intended downstream use.

bioinformatics↗
bioRxiv2026

Decomposing scanner batch effects and their impact on pathology foundation models

Pathology foundation models (PFMs) are expected to represent the same tissue in the same way on any scanner, but their embeddings still carry scanner signals. Scanner-related image differences and the scanner sensitivity of PFMs have each been described. How tissue-dependent changes in color and spatial frequency relate to shifts in PFM representations, and what this means for correction, is less understood. We imaged 103 slides from 37 normal tissue types on six scanners (PanNormal; 95,780 matched locations) and used the real paired images as targets. We used the public PLISM dataset for external validation. First, we decomposed scanner differences into image properties. Color and spatial frequency changed together, by amounts that depended on tissue type. Second, corrections that matched these measured properties brought images closer to the target more reliably than representations. Pix2Pix reduced the image residual by 69% but increased UNI distance to the target by 27%. The same frequency matching helped one scanner and hurt another. Probes still distinguished corrected from real target embeddings after every correction. Third, changing one frequency band at a time showed that PFMs differ in how strongly each band moves them and whether it moves them toward the target. Across eight PFMs pretrained on whole-slide image tiles, the models that moved least under these small changes were the most robust to real scanner differences (Spearman {rho}=-0.90). Aligning images and aligning representations are therefore different goals. The gap between them depends on scanner, tissue, and model.

bioinformatics↗