Matthew T. Wheeler’s research while affiliated with Stanford Medicine and other places

Background: Individuals with neuromuscular diseases (NMDs) have low physical activity levels and an increased risk of cardiovascular and pulmonary diseases. Respiratory gas kinetics obtained during cardiopulmonary exercise testing (CPET) may provide valuable insights into disease mechanisms and cardiorespiratory fitness in individuals with NMD. Recovery from exercise is an important marker of exercise performance and overall physical health, and impaired recovery is strongly associated with poor health outcomes. This study evaluates recovery metrics in individuals with NMD after performing maximal exertion during CPET. Methods: A total of 34 individuals with NMD and 15 healthy volunteers were recruited for the study. CPET was performed using a wearable metabolic system and a wheelchair-accessible total body trainer to peak exertion. Recovery metrics assessed were (i) the time to reach 50% O2 recovery compared with peak exercise and (ii) the ratios of ventilation and respiratory gases between peak exercise and the highest values observed during recovery (overshoot). Results: The NMD group had a significantly longer time to reach 50% O2 recovery (T1/2 VO2: 105 ± 43.4 vs. 76 ± 36.4 s, p = 0.02), lower respiratory overshoot (17.1 ± 13.0% vs. 28.8 ± 9.03%), and lower ventilation/VO2 (31.9 ± 28.3 vs. 52.2 ± 23.5) compared to the control group. Conclusions: This study observes significantly impaired recovery metrics following peak exercise in individuals with NMD compared to controls. These insights may improve the understanding of exercise recovery and mechanics, thus improving prognostication and optimizing exercise prescriptions for individuals with NMD.

Biomedical research underpins progress in our understanding of human health and disease, drug discovery, and clinical care. However, with the growth of complex lab experiments, large datasets, many analytical tools, and expansive literature, biomedical research is increasingly constrained by repetitive and fragmented workflows that slow discovery and limit innovation, underscoring the need for a fundamentally new way to scale scientific expertise. Here, we introduce Biomni, a general-purpose biomedical AI agent designed to autonomously execute a wide spectrum of research tasks across diverse biomedical subfields. To systematically map the biomedical action space, Biomni first employs an action discovery agent to create the first unified agentic environment -- mining essential tools, databases, and protocols from tens of thousands of publications across 25 biomedical domains. Built on this foundation, Biomni features a generalist agentic architecture that integrates large language model (LLM) reasoning with retrieval-augmented planning and code-based execution, enabling it to dynamically compose and carry out complex biomedical workflows -- entirely without relying on predefined templates or rigid task flows. Systematic benchmarking demonstrates that Biomni achieves strong generalization across heterogeneous biomedical tasks -- including causal gene prioritization, drug repurposing, rare disease diagnosis, microbiome analysis, and molecular cloning -- without any task-specific prompt tuning. Real-world case studies further showcase Biomni's ability to interpret complex, multi-modal biomedical datasets and autonomously generate experimentally testable protocols. Biomni envisions a future where virtual AI biologists operate alongside and augment human scientists to dramatically enhance research productivity, clinical insight, and healthcare. Biomni is ready to use at https://biomni.stanford.edu, and we invite scientists to explore its capabilities, stress-test its limits, and co-create the next era of biomedical discoveries.

Background: An estimated 1 in 500 people live with hypertrophic cardiomyopathy (HCM), a disease for which genetic diagnosis can identify family members at risk, and increasingly guide therapy. Mutations in the myosin binding protein C3 (MYBPC3) gene account for a significant proportion of HCM cases. However, many of these variants are classified as variants of uncertain significance (VUS), complicating clinical decision-making. Scalable methods for variant interpretation in disease-specific cell types are crucial for understanding variant impact and uncovering disease mechanisms. Methods: We developed a scaled multidimensional mapping strategy to evaluate the functional impact of variants across a critical domain of MYBPC3. We incorporate saturation base editing at the native MYBPC3 locus, a long-read RNA sequencing-enabled assay of variant splice effects, and measurements of HCM-relevant phenotypes, including MYBPC3 abundance, hypertrophic signaling, and ubiquitin-proteasome function in human induced pluripotent stem cell-derived cardiomyocytes (iPSC-CMs). Results: Our multidimensional mapping strategy enabled high-resolution functional analysis of MYBPC3 variants in iPSC-CMs. Targeted transient base editing generated a comprehensive variant library at the native locus, capturing diverse variant effects on cellular HCM-relevant phenotypes. Our massively parallel splicing assay identified novel splice-disrupting variants. Integration of functional assays revealed that decreased MYBPC3 abundance is a key driver of HCM-related phenotypes. In parallel, downregulation of protein degradation was observed as a compensatory response to MYBPC3 loss of function, and novel disease mechanisms were identified for missense variants near a critical binding domain, underscoring their contribution to pathogenesis. Bayesian estimates of variant effects enable the reclassification of clinical variants. Conclusions: This work provides a platform for extending genome engineering in iPSCs to multiplexed assays of variant effects across diverse disease-relevant cellular phenotypes, enhancing the understanding of variant pathogenicity and uncovering novel biological mechanisms that could inform therapeutic strategies.

BACKGROUND In hypertrophic cardiomyopathy (HCM), the mechanisms through which pathogenic sarcomere variants (G+) lead to left ventricular hypertrophy (LVH) are not understood. METHODS VANISH (Valsartan for Attenuating Disease Evolution in Early Sarcomeric Hypertrophic Cardiomyopathy) was a multicenter, double-blind, placebo-controlled randomized trial testing valsartan’s ability to attenuate phenotypic progression in early sarcomeric (G+LVH+) and subclinical HCM (G+LVH-). The outcome was a composite Z score reflecting cardiac remodeling from baseline to year 2 (end of study). Baseline and year 2 blood samples were used to quantify 276 proteins using a proximity extension assay (Olink, Sweden). We explored relative differences in protein abundance between early and subclinical HCM at baseline. In addition, we compared proteomic changes between baseline and year 2 in subclinical HCM participants who experienced phenotypic conversion to early HCM (converters) versus nonconverters; early HCM participants receiving valsartan versus placebo; and in association with changes in Z score. Comparisons were made using t test/Mann-Whitney U test, linear mixed models, and generalized linear models, correcting for multiple testing. RESULTS Circulating proteins were analyzed in 192 participants (32 subclinical and 160 early HCM [81 allocated to valsartan]). NT-proBNP (N-terminal pro-B-type natriuretic peptide) differentiated early from subclinical HCM and tracked with phenotypic progression in early HCM (1-unit worsening in Z score associated with a 27% increase in NT-proBNP [95% CI, 17–37%]). Some extracellular matrix remodeling proteins showed higher abundance (tissue-type plasminogen activator) in early compared with subclinical HCM or tracked with disease progression (decorin) in early HCM. Growth factors had higher relative abundance in early HCM (fibroblast growth factor-21). While no individual protein was able to distinguish converters from nonconverters, multiprotein the panels lipocalin 2, lectin-like oxidized low-density lipoprotein receptor 1, and either NT-proBNP or interleukin-17 receptor A, could distinguish these groups. CONCLUSIONS NT-proBNP was the most robust protein to track progression. Studying pathways involving growth factors and extracellular matrix remodeling may yield additional insights into mechanisms behind disease progression. REGISTRATION URL: https://www.clinicaltrials.gov ; Unique identifier: NCT01912534.

Background: Individuals with neuromuscular diseases (NMD) have low physical activity levels and an increased risk of cardiovascular and pulmonary diseases. Respiratory gas kinetics obtained during Cardiopulmonary Exercise Testing (CPET) may provide valuable insights into disease mechanisms and cardiorespiratory fitness in individuals with NMD. Recovery from exercise is an important marker of exercise performance and overall physical health, and impaired recovery is strongly associated with poor health outcomes. This study evaluates recovery metrics in individuals with NMD after performing maximal exertion during CPET. Methods: 34 individuals with NMD and 15 healthy volunteers were recruited for the study. CPET was performed using a wearable metabolic system and a wheelchair-accessible total body trainer to peak exertion. Recovery metrics assessed were (i) the time to reach 50% O2 recovery compared with peak exercise and (ii) the ratios of ventilation and respiratory gases between peak exercise and the highest values observed during recovery (overshoot). Results: The NMD group had a significantly longer time to reach 50% O2 recovery (T1/2 VO2: 105±43.4 vs. 76±36.4 seconds, p=0.02), lower respiratory overshoot (17.1±13.0% vs 28.8±9.03%), and lower ventilation/VO2 (31.9±28.3 vs. 52.2±23.5) compared to the control group. Conclusion: This study observes significantly impaired recovery metrics following peak exercise in individuals with NMD compared to controls. These insights may improve the understanding of exercise recovery and mechanics, thus improving prognostication and optimizing exercise prescriptions for individuals with NMD.

Purpose Whole-exome sequencing (WES) and whole-genome sequencing (WGS) are increasingly used as standard genetic tests to identify the diagnostic variants in rare disease cases. However, prioritizing these variants to reduce the time and burden of manual interpretation by clinical teams remains a significant challenge. The Exomiser/Genomiser software suite is the most widely adopted open-source software for prioritizing coding and non-coding variants. Despite its ubiquitous use, limited data-driven guidelines currently exist to optimize its performance for diagnostic variant prioritization. Based on detailed analyses of Undiagnosed Diseases Network (UDN) probands, this study presents optimized parameters and practical recommendations for deploying the Exomiser and Genomiser tools. We also highlight scenarios where diagnostic variants may be missed and propose alternative workflows to improve diagnostic success in such complex cases. Methods We analyzed 386 diagnosed probands from the UDN, including cases with coding and non-coding diagnostic variants. We systematically evaluated how tool performance was affected by key parameters, including gene:phenotype association data, variant pathogenicity predictors, phenotype term quality and quantity, and the inclusion and accuracy of family variant data. Results Parameter optimization significantly improved Exomiser’s performance over default parameters. For WGS data, the percentage of coding diagnostic variants ranked within the top ten candidates increased from 49.7% to 85.5%, and for WES, from 67.3% to 88.2%. For non-coding variants prioritized with Genomiser, the top ten rankings improved from 15.0% to 40.0%. We also explored refinement strategies for Exomiser outputs, including using p -value thresholds and flagging genes that are frequently ranked in the top 30 candidates but rarely associated with diagnoses. Conclusion This study provides an evidence-based framework for variant prioritization in WES and WGS data using Exomiser and Genomiser. These recommendations have been implemented in the Mosaic platform to support the ongoing analysis of undiagnosed UDN participants and provide efficient, scalable reanalysis to improve diagnostic yield. Our work also highlights the importance of tracking solved cases and diagnostic variants that can be used to benchmark bioinformatics tools.