bioRxiv Science⌕ Search

Biology subjects

Zhao, W.

Publications and source records attributed to Zhao, W..

At least 19 recordsLinked to original sources

Systematic analysis of genetic vulnerabilities caused by chromosome loss events in cancer

Chromosome loss events are common across tumor types but rarely occur in normal tissue, raising the possibility that these alterations could be exploited as cancer-specific therapeutic vulnerabilities. Here, we integrate genome-scale CRISPR screening data with chromosome-level copy number profiles from 1,150 cancer cell lines and identify 389 dependencies that are selectively associated with individual chromosome loss events. The most significant vulnerabilities uncovered by our analysis are paralogous gene pairs, in which the loss of a paralog on an aneuploid chromosome establishes a unique dependency on its redundant paralog encoded elsewhere in the genome. We experimentally validate two paralog dependencies: loss of chromosome 8p, which encodes the phosphatase PPP2CB, creates a specific dependency on its paralog PPP2CA, while loss of the Y chromosome, which encodes the helicase DDX3Y, creates a specific dependency on its paralog DDX3X. Systematic benchmarking of CRISPR screening data indicates that the selective dependency on DDX3X in loss-of-Y cells exceeds other published aneuploidy-associated vulnerabilities and matches or exceeds most clinically-validated biomarker-target relationships. Furthermore, we found that the acute degradation of DDX3X was sufficient to trigger apoptosis in vitro and tumor regressions in mice. Collectively, these findings establish chromosome loss as a recurrent source of biomarker-defined therapeutic targets, identify paralog insufficiency as a general vulnerability in aneuploid cells, and provide in vivo proof-of-concept for targeting DDX3X in cancers with Y chromosome loss.

cancer biology↗

NeCS-Flow: Necrosis-Guided Contrast-Enhanced MRI Synthesis with Rectified-Flow Refinement

Background: Gadolinium-based contrast agents raise concerns about tissue deposition, adverse reactions, and increased costs. Purpose: To develop and validate a deep learning pipeline for gadolinium-free post-contrast T1w brain MRI synthesis incorporating necrotic scaling and Posterior-Mean Rectified Flow (PMRF) refinement, and to evaluate diagnostic utility through a blinded multi-reader study. Study Type: Retrospective. Population: 454 patients with adult-type gliomas (grade 2-4) from four publicly available datasets. Training used 374 patients; internal testing used 60 held-out patients; external validation used 20 patients. Field Strength/Sequence: 1.5-3.0 T; pre-contrast T1w, T2w, and FLAIR as inputs; post-contrast T1w as synthesis target. Assessment: A three-stage pipeline comprising nnU-Net Necrotic core segmentation, necrotic-scaling-guided U-Net Contrast Synthesis, and PMRF perceptual refinement (NeCS-Flow) was evaluated using whole-brain and tumor-region similarity metrics, enhancing tumor Dice scores, perceptual quality metrics, and a blinded three-reader study assessing diagnostic accuracy, enhancement detection, image quality, and real-versus-synthetic discriminability. Statistical Tests: Paired t-tests and two one-sided tests for model comparison; Fisher's exact test and two-sample t-tests for reader study comparisons, with p < 0.05 considered significant. Results: Necrotic scaling consistently improved real versus synthetic enhancing tumor Dice internally (HGG: 0.793 to 0.816) and externally (HGG: 0.465 to 0.612). NeCS-Flow outperformed two benchmark architectures on all quantitative metrics in internal and external testing, although absolute enhancing tumor accuracy declined for all models on the external cohort. PMRF refinement improved perceptual realism across both tumor types, reducing both FID and KID. Readers distinguished synthetic from real images only modestly above chance (overall accuracy 61.7%), and no significant differences between real and synthetic images were found for any clinical metric. Data Conclusion: NeCS-Flow synthesizes perceptually realistic and diagnostically equivalent post-contrast T1w brain MRI, with necrotic scaling improving enhancing tumor delineation and PMRF refinement improving perceptual realism across gliomas. Performance on an external cohort was preserved relative to benchmark methods, though reduced in absolute terms.

bioinformatics↗

Cortex-wide single-cell activity mapping in freely moving mice

How the cortex supports complex, naturalistic behavior can be revealed only in freely moving animals. Although head-mounted microscopes promise access to the neural dynamics underlying such behavior, fundamental trade-offs among field of view, spatial resolution, device weight, and background contamination have prevented cortex-wide single-neuron recording, leaving distributed cortical coordination largely inaccessible. To meet these challenges, we introduce COMET (Cortex-wide Observational Miniature Epifluorescence Technique), a 3.75-g head-mounted mesoscope that integrates an aberration-controlled wide-field optical design, uniform high-efficiency excitation and single-coaxial power-and-data transmission with an end-to-end denoising and signal-extraction pipeline. This enables cortex-wide calcium imaging at single-cell resolution in freely moving mice. COMET captures activity from >15,000 layer 2/3 neurons per 5-min recording over fields of view up to 10 mm in diameter while preserving natural behavior. Benchmarking against benchtop two-photon microscopy demonstrates high spatial concordance of detected neurons and matched event-level calcium dynamics. We apply COMET to social interaction to quantify region-specific engagement and to an auditory trace fear-conditioning task that requires natural movement and behavioral expression, revealing time-ordered changes in regional firing-rate structure and functional connectivity preceding a global conditioned response. COMET provides an open, lightweight hardware-software framework for linking single-neuron activity to cortex-scale network dynamics during freely expressed behaviors under changing internal or external conditions.

neuroscience↗

Engineered Chimeric Proteins Functionally Replace ALS-causing Endogenous FUS and TDP-43 to Improve Neuronal Health

Mislocalization and aggregation of the RNA-binding proteins (RBPs) FUS and TDP-43 are pathological hallmarks of amyotrophic lateral sclerosis (ALS) and frontotemporal dementia (FTD), contributing to combined loss- and gain-of-function toxicity. Here, we present a protein engineering strategy that appends optimized anti-aggregation motifs to FUS and TDP-43, thereby enhancing their nuclear localization and solubility, and enabling functional replacement of disease-associated proteins. We show that these engineered chimeric RBPs reduce levels of aggregation-prone endogenous proteins through autoregulatory mechanisms. In addition, chimeric FUS and TDP-43 recapitulate key nuclear functions, including DNA repair and mRNA processing. Notably, chimeric FUS restores neurite outgrowth in diseased iPSC-derived neurons, while chimeric TDP-43 suppresses cryptic exon inclusion in patient-derived fibroblasts. Collectively, these results demonstrate that engineered chimeric RBPs can rebalance RBP homeostasis by simultaneously addressing protein aggregation and mislocalization, thus mitigating RBP-associated loss- and gain-of-function toxicity, highlighting a generalizable strategy for therapeutic protein replacement in neurodegenerative disease.

neuroscience↗

Mesaconic acid delivery via biomaterials and engineered probiotics to the gut reduces inflammation in Crohn's Diseases mouse models

Mesaconate (MES, an immunosuppressive metabolite) but not itaconate is upregulated in stool of Crohns Disease (CD) patients, albeit below therapeutic levels.. However, colonic MES level is difficult to maintain because of its high absorption rate, limiting its therapeutic potential for CD. Herein, we developed two strategies that elevate MES level and suppress inflammation: EcN-CAD, an engineered probiotic Escherichia coli Nissle 1917 (EcN) that overexpresses cis-aconitate decarboxylase (CAD, produces itaconic acid that is converted to MES); and MES+Zn, a Zinc-MES ionically crosslinked formulation. Mechanistically, MES modulates T cell function by downregulating glycolytic and mitochondrial activity and decrease ROR{gamma}t expression. In preclinical SAMP1/YitFC dexamethasone-withdrawal model and TNFDARE model, MES+Zn decreased gut permeability significantly compared to controls and reduced myeloperoxidase (MPO, an inflammatory marker) 1000-fold compared to controls. EcN-CAD probiotics decreased ROR{gamma}t expression in CD4+ and CD8+ T cells, gut permeability and inflammation in preventive SAMP1/YitFC model, and SAMP1/YitFC dexamethasone-withdrawal model. Overall, MES rewires T cell responses and technologies that enhance MES levels can be effective in gut-related inflammatory diseases.

immunology↗

THE MICROANATOMY OF INTRACRANIAL HUMAN THROMBI BY SINGLE NUCLEI TRANSCRIPTOMICS

Rationale: Thrombi are fragile, heterogeneous, and difficult to study with available methods, yet understanding their molecular composition and local microenvironment in stroke is critical to elucidating stroke pathophysiology and improving patient outcomes. Objective: We used a novel platform for harvesting and analyzing intracranial thrombi from endovascular thrombectomy for large vessel occlusion to define the cellular and transcriptomic landscape of thrombi. Methods and results: We performed single-nuclei RNA sequencing (snRNA-seq) on three human intracranial thrombi to define their molecular composition and microenvironment. Across 5,143 nuclei, we identified 8 distinct cellular populations, including monocytes and T cells. Monocytes displayed transcriptional changes on gene and regulon levels suggestive of transdifferentiation. We integrated our snRNA-seq data with publicly available single cell RNA-sequencing (scRNA-seq) data from asymptomatic and symptomatic atherosclerotic carotid plaques, and scRNA-seq data from healthy brain vasculature. We identified shared transcriptional programs between carotid and thrombi monocytes along dimensions including inflammation and endothelial activation, supporting the notion that thrombi monocytes undergo transdifferentiation in stroke. We also inferred potential ligand-receptor interactions that may occur at the site and time of occlusion. Conclusion: Our novel application of snRNA-seq to thrombectomy specimens has allowed us to uncover the full diversity of cell types present in thrombi, including low-abundance and novel cell populations, and to provide evidence of cellular plasticity and intercellular communication at the site of occlusion. This detailed definition of the cellular populations involved in stroke will facilitate mapping of novel interventional targets with direct functional relevance and facilitate greater personalization of post-reperfusion therapy.

molecular biology↗

Parental Genome Assemblies of Suyunuo1 Reveal Structural Variation Underlying Edible Waxy Maize Evolution, Superior Hybrid Performance and Yield - Flavor Balance

Edible waxy maize is a unique domesticated cereal valued for its superior sensory and nutritional properties, yet high quality gap free genomes remain lacking, hindering the exploration of structural variations (SVs) and domestication related divergence. Here, we assembled two chromosome level, gap free genomes of the elite waxy maize inbred lines Tongxi 5 and Hengbai 522, the parents of the widely cultivated hybrid Suyunuo 1. Comparative genomic analysis revealed substantial genome size variation, prominent megabase scale SVs, and extensive knob repeat expansion in Tongxi 5. We identified widespread gene presence absence variation and hyperdivergent regions enriched in transposable elements, which largely underpin genomic differentiation between waxy and field maize. Population genomic analyses further demonstrated asymmetric introgression from field maize into the two parents, whereas the conserved waxy haplotype supports a shared ancestral origin. Integrative GWAS based on SNPs, InDels and SVs revealed that SVs substantially contribute to phenotypic diversity and mediate the co-regulation of yield traits and flavor-related metabolites. Our results uncover key genomic events underlying post-domestication divergence and highlight the essential roles of SVs in heterosis and flavor-yield balance, providing valuable genomic resources for waxy maize improvement.

plant biology↗

An Excelled Trifluorinated Probe for GPCR Conformational Quantification

G protein-coupled receptors (GPCRs) represent the largest membrane protein family in the human body and the major targets of FDA-approved medications. Unlike some other physiological proteins, GPCRs are highly plastic and often exist as collections of conformational ensembles in the environment to exert their function. Therefore, understanding how a GPCR adopts different conformational ensembles in response to ligands and signaling partners is essential for drug development. 19F-NMR has emerged as one of the most promising tools in this regard; however, a well-established 19F-labeling system is still lacking, including an effective 19F probe that meets both environmental susceptibility and NMR sensitivity requirements, even though progress has been made dramatically in recent years. Thus, this manuscript reports a novel conformational probe (4-(bromomethyl)-ethoxy-2-(trifluoromethyl)benzene, BTMB) that offers paramount benefits for future GPCR conformational dynamics studies. To rationalize the mechanism behind BTMB's superiority for future probe development, we performed MD simulations, referenced to the current state-of-the-art probe, BTFMA. Our study suggests that the spatial proximity of aromatic residues significantly influences the chemical shift dispersion and linewidths of related 19F-probe-profiled conformational states, as well as the rigidity and compactness of -CF3 in the target protein.

biophysics↗

An Artificial Intelligence Model for Longitudinal Assessment of TCR Repertoires in SARS-CoV-2 Vaccine Recipients

T cells play a crucial role in reducing disease severity during SARS-CoV-2 infection and in shaping long-term immune memory. However, the precise molecular immune responses, particularly involving T-cell receptor (TCR) repertoire changes after full vaccination, and the use of TCR analysis to evaluate vaccine efficacy, remain incompletely understood. In this study, we developed the DeepAir-Cov19 model (AUC=0.94), a large language model tailored to identify SARS-CoV-2-specific TCRs, and observed significant differences in the TCR profiles between antibody-negative and antibody-positive populations before and after vaccination, indicating that the immune status of pre-vaccine recipients can directly assess the efficacy of vaccination. Notably, SARS-CoV-2-specific TCRs expanded to peak levels after the second dose and remained detectable in most subjects up to 10 months post-vaccination. Meanwhile, we also identified three specific V genes, 20 V-J combinations, and 3 epitopes associated with these responses. Finally, by leveraging vaccine-specific TCRs as novel biomarkers, we developed a vaccine efficacy model that predicts antibody levels with a mean AUC of 0.96. These findings underscore the accuracy of the large model in predicting SARS-CoV-2-specific TCRs and reveal a strong correlation between SARS-CoV-2-specific TCR responses and antibody levels. This highlights the complementary and synergistic roles of T cells and antibodies in providing vaccine-mediated protection. Our results offer valuable insights into the longitudinal dynamics of SARS-CoV-2-specific TCRs and illustrate the potential for developing potency evaluation models based on these TCR insights.

bioinformatics↗

Protein-driven colloid-osmotic pressure controls nuclear size, organization, and function

The size of the cell nucleus is tightly controlled and changes in nuclear size correlate with altered nuclear function during development, cell differentiation and senescence. How nuclear size is regulated and whether changes thereof are functionally relevant remains unclear. Here, we demonstrate that nuclear size is determined by the osmotic pressure exerted by proteins in yeast, human cells and frog egg extracts. The biophysical model we establish solves the long-standing question of how the nuclear-to-cytoplasmic ratio is regulated and maintained. Furthermore, altering protein-driven osmotic balance modulates the physical properties of the nucleus, with a direct effect on chromatin organization and gene expression: Nuclear enlargement causes the dissolution of heterochromatic structures and derepression of subtelomeres and transposons, while simultaneously down-regulating highly expressed housekeeping genes. Importantly, these global transcriptional patterns closely mimic the gene expression changes that occur as yeast, human, and drosophila cells enlarge. Importantly, artificially forcing nuclear compression is sufficient to reverse these size-associated expression changes. Together, our findings provide a quantitative, mechanistic explanation for the coupling between the size of the nucleus and the cell, and they establish nuclear size as a modulator of chromatin organization and gene expression.

cell biology↗

Comprehensive characterization of genomic, transcriptomic and epigenomic artifacts introduced in formalin-fixed, paraffin-embedded tissues.

Genomic, transcriptomic and epigenomic characterization has accelerated the discovery of clinically-relevant alterations in cancer, predominantly using fresh frozen (FF) specimens. However, clinical molecular pathology laboratories prefer formalin-fixed paraffin-embedded (FFPE) methods, known to introduce artifacts at the nucleic acid level, over fresh frozen methods. Extending the multi-platform analysis to FFPE specimens for comprehensive clinical molecular diagnosis requires a thorough understanding of the consequence of formalin-fixation. We present a detailed multi-platform characterization of FFPE preservation using paired FF specimens as the 'gold standard'. DNA and RNA were obtained from 38 patients across 6 cancer types using a FFPE optimized co-isolation. The impact of FFPE on exome sequencing was dependent on filtering, where a minimum coverage or supporting read filter can mitigate FFPE-specific false positives. Copy number alterations, MSI assessment, mutational signatures, and DNA methylation were comparable between FFPE and FF. FFPE biases in RNA expression can be overcome when using biology-relevant genes and we describe a novel consequence of FFPE on miRNA species diversity. Collectively, this data provides a broad view of FFPE artifact and offers best practices for overcome these biases.

bioinformatics↗

Single-atom inhibition of oncogenic drivers through cysteine coordination

Small-molecule inhibitors rely on molecular recognition within suitable binding pockets, leaving many disease-associated proteins difficult to target. Here, we introduce the concept of a single-atom inhibitor in which gold (Au) engages critical cysteine residues of oncogenic drivers to suppress their activity. We used an AI-assisted few-shot learning approach to identify EGFR-targeting peptides for in vivo Au delivery and showed that the lead candidate, 10714, promoted Au accumulation in EGFR-expressing cells and tumors. In vivo, Au exploited its intrinsic affinity for cysteine to inhibit two structurally distinct oncogenic proteins, engaging Cys797 in EGFR T790M and the mutation-derived Cys12 in KRAS G12C adjacent to their respective nucleotide-binding pockets. Structural and computational analyses supported stabilization of inactive nucleotide-bound states, while mutation of these cysteine residues abrogated Au-mediated inhibition. 10714-Au consequently suppressed oncogenic signaling, reduced non-small-cell lung cancer cell viability, and inhibited tumor growth in EGFR- and KRAS-mutant xenograft models and patient-derived organoids. These findings establish proof of principle for single-atom inhibition across structurally distinct oncogenic drivers and suggest that localized atomic coordination could provide an alternative mode of target engagement to conventional pocket-dependent inhibition.

molecular biology↗

NACraft: Programmatic nucleic-acid aptamer design via all-atom structure-model feedback

Protein-nucleic-acid interactions underpin diverse biological processes and provide a basis for molecular sensing, regulation and therapeutic intervention. However, the coupled dependence of aptamer function on nucleotide sequence, three-dimensional folding and target binding makes rational RNA and DNA binder design challenging. Here we present NACraft, a training-free and programmatic framework for all-atom nucleic-acid aptamer design based on backpropagation through structure-model feedback. By composing binding, sequence-similarity and anti-binding constraints, NACraft supports de novo generation, similarity-guided sampling and target-selective design within a unified optimization framework, without task-specific training or fine-tuning. Computational experiments showed that NACraft generated high-confidence candidates de novo across diverse protein targets, with further improvements achieved through similarity-guided design for both RNA and DNA complexes. Its target-selective design capability was further validated in silico, with 69.44% of paired candidates generated to favour the positive target EGFR over the off-target HER2. Under matched independent AlphaFold3 evaluation, NACraft achieved better performance than ODesign in 10 of 11 NA-12 targets and 17 of 20 protein target-length settings. Together, these results demonstrate the effectiveness and versatility of NACraft and extend structure-model hallucination toward programmatic nucleic-acid aptamer design. Codehttps://github.com/OTEAM-AI4S/NACraft

bioinformatics↗

Diatoms maintain metabolic activity in deep ocean twilight zones

Phytoplankton are primarily confined to the sunlit epipelagic zone, yet intact algal cells persist in the oceans mesopelagic twilight zone where light is minimal. However, their survival strategies in the deeper ocean remain unknown. Here, we isolate Chaetoceros sp. DS1 from 1000 m depth and reveal its survival mechanisms under prolonged darkness and high pressure, including membrane lipid remodeling and enhanced antioxidant defenses. Importantly, DS1 maintains a latent photosynthetic reserve that activates upon exposure to dim blue light and higher pressure typical of the twilight zone, enabling photosynthesis and nutrient uptake. It exhibits metabolic flexibility by utilizing stored carbon in darkness and assimilating organic carbon under dim light. Global meta-omics data show widespread, transcriptionally active Chaetoceros-like diatoms in the twilight zone, expressing blue-light sensors and protein synthesis genes. These findings identify the ocean twilight zone as a previously overlooked niche for metabolically active phytoplankton, expanding understanding of deep-ocean microbial ecology and carbon cycling.

ecology↗

Characterizing Functional Clusters of V4 Neurons in Digital Twins

Neurons in primate visual cortical area V4 display tuning for multiple visual features, including color, shape, texture, and depth. Whether and how these neurons are organized into functional architectures remains largely unknown. Using two-photon calcium imaging in anesthetized macaques, we recorded responses of hundreds of V4 neurons to natural images and used these data to train deep convolutional neural network models (digital twins), obtaining synthetic images (SI) that maximized neuronal responses. Based on their SIs, these neurons clustered into classes that spatially matched the orientation, color, and curvature maps from intrinsic signal imaging. Furthermore, lesion study in digital twins revealed different integration rules for different neuron classes. Thus, digital twins of V4 neurons can be applied as a promising tool to characterize neurons fundamental features, which underlie the functional clustering in this area. HighlightsO_LINeurons in V4 are examined with two-photon calcium imaging and DCNN modeling (digital twins) C_LIO_LISynthetic images derived from neurons digital twins reveal distinct neuron groups C_LIO_LIThese groups match functional types defined by intrinsic signal optical imaging C_LIO_LILesion study in digital twins demonstrates neural mechanisms underlying feature tuning C_LI

neuroscience↗

ZDHHC14-Mediated Palmitoylation of TBK1 Promotes Pathological Cardiac Hypertrophy via Type I Interferon Signaling Activation

Pathological cardiac hypertrophy represents a common maladaptive response to cardiovascular stress and constitutes a major harbinger of heart failure. Although S-palmitoylation--a reversible post-translational modification--critically governs protein localization, trafficking, and stability, its involvement in cardiac hypertrophy remains poorly characterized. In this study, we aimed to explore the role and regulatory mechanism of a palmitoyltransferase, zinc finger DHHC-type palmitoyltransferase 14 (ZDHHC14), in cardiac hypertrophy. We found that ZDHHC14 was significantly upregulated in cardiac hypertrophy tissues from both human patients and mouse models. Cardiomyocyte-specific ZDHHC14 knockdown ameliorated transverse aortic constriction (TAC)-induced cardiac hypertrophy and dysfunction in male mice, whereas cardiac-specific ZDHHC14 overexpression via AAV9 exacerbated these pathological phenotypes. Mechanistically, TANK-binding kinase 1 (TBK1) was identified as a novel substrate of ZDHHC14 through interactomic screening. ZDHHC14 catalyzed TBK1 palmitoylation at cysteine 267, which in turn facilitated TBK1 phosphorylation and subsequent activation of type I interferon (IFN-I) signaling, ultimately promoting cardiac hypertrophy. Importantly, our findings demonstrate that the TBK1-C267S mutation rectifies ZDHHC14 overexpression-induced exacerbation of cardiac hypertrophy. This study illustrated a ZDHHC14-TBK1-IFN-I axis in regulating cardiac hypertrophy, which may provide a potential therapeutic target to ameliorate pathological cardiac hypertrophy.

molecular biology↗

Time-dependent BMP4 signaling directs lineage specification in human mesoderm

Human pluripotent stem cells (hPSCs) provide a powerful platform for modeling early human embryonic development. Here, we investigate the mechanisms underlying mesodermal heterogeneity using a minimal directed differentiation system that simultaneously generates paraxial (PXM), intermediate (IM) and lateral plate mesoderm (LPM) populations. Single-cell RNA sequencing across defined time points during hPSC differentiation revealed a temporal sequence of lineage specification with LPM emerging first, followed by PXM and IM differentiation. Ligand-receptor and differential gene expression analyses identified BMP4 as a key regulator enriched in LPM-associated clusters versus mesoderm progenitors (MPs) that hold PXM and IM precursors. Whereas LPM cells cluster with an early BMP4 signal, IM clusters are associated with later BMP4. Moreover, these early and late BMP4 signals regulate this lineage specification potentially through distinct downstream pathways. Leveraging this insight, we established a stepwise protocol combining early BMP inhibition with subsequent BMP4 supplementation, suppressing initial LPM fate to efficiently induce IM from a mixed MP population. Longer culture of these selective IM progenitors promotes more mature nephrogenesis. Moreover, we demonstrate that during early differentiation high levels of BMP4 can still redirect MPs to more lateroventral fates, illustrating a degree of plasticity within the mesoderm lineage. Together, our results define a temporal framework for BMP4 signaling in mesoderm fate determination and provide a strategy for selective mesoderm differentiation from hPSCs. HIGHLIGHTSO_LIDevelopment of a minimal 2D differentiation platform allows for heterogenous mesoderm formation. C_LIO_LITemporal BMP4 signaling differentially directs mesoderm fates, with early exposure favoring LPM and late exposure promoting IM identity. C_LIO_LILPM cells arise first while later mesoderm progenitors hold both IM and PXM-fated cells. C_LIO_LISequential BMP modulation promotes IM and enhances nephrogenesis. C_LI

developmental biology↗

Performance Attribution in pLM-Based Biological Relation Prediction

Protein language models (pLMs) have enabled strong performance in biological relation-prediction benchmarks, yet aggregate predictive metrics do not reveal which sources of information support that performance. We examined three pLM-based case studies--MetaESI, Deep-GNHV, and SAGEPhos--using frozen pLM-derived full-input baselines, component-restricted controls, train-derived endpoint or site priors, and selector-matched random restrictions. Generic tabular learners trained on frozen pLM-derived inputs achieved strong task-specific benchmark references, reaching AUROC/AUPRC values of 0.827/0.703 for the MetaESI-derived full-input rerun, 0.922/0.708 for DeepGNHV, and 0.896/0.893 for SAGEPhos. Across tasks, substantial predictive signal remained available from restricted endpoint, site/window, motif, or train-derived prior features. In the MetaESI-derived row-level evaluation, a self-label-excluding endpoint-prior control reached AUROC/AUPRC of 0.846/0.676 without sequence embeddings, numerically close to the frozen full-input reference. In a separate frozen-pooling experiment, GARD-selected positions did not outperform count-matched random token pooling. Endpoint-cold diagnostics further showed substantial performance degradation when one or both endpoint types were unseen during training. These findings do not diagnose leakage, imply memorization, exclude biological learning, or invalidate the evaluated models. Rather, they show that strong benchmark performance can reflect multiple overlapping information sources and that aggregate scores alone cannot establish pair-specific learning, the incremental value of task-specific architectures, selector-specific information content, or mechanistic biological interpretation. We therefore propose that such claims require matched attribution controls, representation-matched baselines, and evaluation protocols aligned with the intended generalization setting.

bioinformatics↗