Scott Sommerfeldt Elected ASA President

Artificial intelligence foundation models are increasingly deployed for prostate cancer Gleason grading, where GP3/GP4 distinction directly impacts treatment decisions (active surveillance vs. intervention). However, these models may achieve high validation accuracy by learning specimen-specific artifacts rather than generalizable biological features, limiting real-world clinical utility. We introduce PANDA-PLUS-Bench, a curated benchmark dataset derived from expertly annotated prostate biopsies designed specifically to quantify this failure mode. The benchmark comprises nine carefully selected whole slide images from nine unique patients containing diverse Gleason patterns, with non-overlapping tissue patches extracted at both 512 × 512 and 224 × 224-pixel resolutions across eight augmentation conditions. Using this benchmark, we evaluate seven foundation models (Virchow, Virchow2, UNI, UNI2, Phikon, Phikon-v2, and HistoEncoder) on their ability to separate biological signals from slide-level confounders. Our results reveal substantial variation in robustness across models: the Virchow models achieved the lowest slide-level encoding among large-scale models (slide ID accuracy: 80.7–81.0%), yet Virchow2 exhibited the lowest cross-slide accuracy (47.2%). HistoEncoder, trained specifically on prostate tissue, demonstrated the highest cross-slide accuracy (59.7%) and the strongest slide-level encoding (slide ID accuracy: 90.3%), suggesting tissue-specific training may enhance both biological feature capture and slide-specific signatures. All models exhibited measurable within-slide vs. cross-slide accuracy gaps, though the magnitude varied from 19.9 percentage points (HistoEncoder) to 26.9 percentage points (Phikon). We provide an open-source Google Colab notebook enabling researchers to evaluate additional foundation models against our benchmark using standardized metrics. PANDA-PLUS-Bench addresses a critical gap in foundation model evaluation by providing a purpose-built resource for robustness assessment in the clinically important context of Gleason grading.

Pyrochlore magnets of the form 𝑅₂𝐵₂O₇, in which rare-earth ions on the 𝑅 site form a three-dimensional network of corner-sharing tetrahedra, provide a canonical setting for geometrical frustration. Ho-based pyrochlores host a dipolar spin-ice ground state, characterized by Ising moments constrained by the ice rules and elementary excitations analogous to magnetic monopoles. Here we examine how controlled chemical disorder influences this state by introducing site mixing on the nonmagnetic 𝐵 site in two compounds. Ho₂GaSbO₇ contains only Ga³⁺/Sb⁵⁺ charge disorder, whereas Ho₂ScSbO₇ exhibits both charge and substantial size disorder arising from the large ionic-radius mismatch between Sc³⁺ and Sb⁵⁺. Although both materials retain the pyrochlore structure, neutron-scattering measurements reveal a reduced correlation length for the 𝑅/𝐵-site cation ordering and enhanced local structural distortions in Ho₂ScSbO₇. Despite these structural differences, bulk thermodynamic measurements and magnetic diffuse scattering demonstrate that both systems exhibit the defining signatures of a dipolar spin-ice state. Low-energy inelastic neutron spectroscopy further uncovers broad magnetic excitations that develop within the dipolar spin-ice regime, a feature absent in pristine Ho pyrochlores and indicative of disorder-induced splitting of the non-Kramers ground-state doublet. Together, these results show that controlled disorder generates tunable transverse-field-driven quantum fluctuations in Ho-based pyrochlores, although the dipolar spin-ice state is remarkably robust to this disorder.

Borna disease virus 1 (BoDV-1) is a non-segmented negative-strand (NNS) RNA virus that uniquely replicates in the nucleus of mammalian host cells, in contrast to most NNS RNA viruses that replicate in the cytoplasm. The mechanisms underlying nuclear replication of BoDV-1 and related bornaviruses with their RNA-dependent RNA polymerase (RdRp) complexes remain poorly understood. Here, we report the 2.8 Å cryo-EM structure of the BoDV-1 RdRp complex, comprising the large (L) protein and tetrameric phosphoprotein (P). The L protein features an N-terminal superdomain containing the RdRp and GDP polyribonucleotidyltransferase (PRNTase, mRNA-capping enzyme) domains, along with three C-terminal appendages, including a methyltransferase-like domain. The RdRp initiates de novo RNA synthesis internally at the genomic promoter, producing 5′-triphosphorylated transcripts corresponding to the 5′ end of the anti-genome. P interacts with the fingers RdRp subdomain of L. Structure-guided mutagenesis shows that the residues involved in the L–P interaction are essential for efficient transcription initiation and, consequently, for viral gene expression. A flexible loop within the PRNTase domain, analogous to the rhabdovirus priming-capping loop, appears critical for transcription initiation. These findings provide the structural and functional insights into the BoDV-1 RdRp and support a shared evolutionary origin between nuclear and cytoplasmic NNS RNA viruses.

Glycine receptors (GlyRs), pentameric ligand-gated ion channels (pLGICs), mediate sensory and motor functions. GlyR functional states are well characterized; however, structural details of transitions between states remain undefined. Here, we determined cryo–electron microscopy structures of GlyRα1β (with gephyrin E-domain) at varying concentrations of ivermectin, a transmembrane domain (TMD) allosteric agonist, and at saturating concentrations of strychnine, a competitive antagonist at the extracellular domain (ECD). Electrophysiology shows that ivermectin activates GlyR even with strychnine present. Structures with both ligands reveal intermediate states featuring a desensitized TMD and an ECD between closed and desensitized conformations, providing insights into domain cooperativity and ligand efficacy. Molecular dynamics simulations show how ivermectin affects strychnine dynamics. These data support a model where ivermectin activates GlyRs through a concerted and near-symmetric TMD mechanism, whereas allosteric ECD motions are graded and spatially heterogeneous. These findings reveal unanticipated features of GlyR gating and establish principles of allosteric modulation applicable to pLGICs.

Glycine receptors (GlyRs) mediate inhibitory neurotransmission in the central nervous system. The GlyRα2 subtype contributes to critical neural circuitry in early neurodevelopment and is also found in adults. GlyRα2 dysfunctions are implicated in neurodevelopmental disorders, including autism, epilepsy, and cognitive delays. GlyRα2 functional properties and pharmacology are distinct from GlyRα1, but the structural basis for these differences remains poorly defined. Here, we report cryo-electron microscopy structures of full-length, human GlyRα2 reconstituted in peptidiscs captured in multiple conformational states. In addition to symmetric resting and desensitized states, we resolved an asymmetric open state, previously observed only in heteromeric GlyRs. This suggests that asymmetry is intrinsic to GlyRα2, independent of β-subunit incorporation. Furthermore, we identified distinct conformations of GlyRα2 with the pore-blocker picrotoxin, providing new insights into allosteric interactions. These findings uncover the structural basis of GlyRα2 function, providing a foundation for understanding its role in development and in GlyRα2-associated disorders.

Despite significant advances in artificial intelligence (AI) algorithms for prostate cancer detection from whole slide images, the clinical applicability of these models remains limited. Variability in inter- and intrapathological grading, low generalizability of training datasets, and insufficient annotation precision restrict the performance of downstream models. This article introduces a novel Bayesian framework that addresses these challenges by generating pixel-wise posterior distributions, thereby providing a probabilistic output that enables the simulation of a panel consensus and enabling seamless integration of new data and models as they become available. The framework is demonstrated by integrating a Bayesian prior with a trained AI model to produce a per-pixel distribution of Gleason patterns. It is shown that using this distribution of Gleason patterns rather than a ground-truth label can improve model applicability, mitigate errors, and highlight areas of interest for pathologists. Furthermore, we present a high-quality, hand-curated dataset of prostate histopathological images annotated at the gland level by trained premedical students and verified by an expert pathologist. We highlight the potential of this adaptive and uncertainty-aware framework for developing clinically deployable AI tools that can support pathologists in accurate prostate cancer grading, improve diagnostic accuracy, and create positive patient outcomes. This work is presented as an early-stage, proof-of-concept study; the framework has not been validated for clinical use and is not intended for diagnostic deployment in its current form.