AKAkihiro Kitani

Akihiro Kitani

*This is an algorithmically generated author record based on publicly available information.

Current Organizations / Institutes

The organizations/institutes from the scholar's latest article


Akihiro Kitani has a total of 15 co-authors. Here are the top co-authors by number of shared publications:


Latest Publications

  • Open Access

    CATEomics: model selection and ensembling for heterogeneous treatment effect estimation from high-dimensional omics data

    by Takeru Furuta, Koichi Himori, Bingyuan Zhang, Akihiro Kitani, Yusuke Matsui

    CATEomics addresses the lack of a standardized and reproducible process for choosing among conditional average treatment effect (CATE) estimators in high-dimensional omics studies, where their estimation errors cannot be evaluated directly from observed outcomes. The R package accepts a SummarizedExperiment, applies common preprocessing and data partitions across R and Python estimators, compares their out-of-fold estimates using proxy losses, supports model selection and ensembling, and returns the resulting CATE estimates to the original object for downstream omics analysis. Availability and implementation: CATEomics is implemented in R and is freely available under the MIT license at https://github.com/matsui-lab/CATEomics. The software version described here (v0.99.0) is archived on Zenodo at https://doi.org/10.5281/zenodo.24189754. A vignette illustrates the complete workflow from a SummarizedExperiment to downstream analysis.

  • Open Access

    When can site-specific glycoform compositions resolve biosynthetic rates?

    by Yusuke Matsui, Koichi Himori, Bingyuan Zhang, Akihiro Kitani

    Mass spectrometry measures site-specific glycoform compositions, but when these measurements can resolve upstream biosynthetic rates remains unclear. We assess whether small changes in effective rates, each representing the processing capacity of a group of biosynthetic reactions, can be distinguished near specified baseline rates with secretion held fixed. Because proportions sum to one, m jointly normalised glycoform classes provide at most m − 1 independent quantities. Distinguishing changes in k rates therefore requires at least k + 1 classes, although this count alone does not ensure identifiability. In a canonical N-glycan network with eight rates, exhaustive evaluation of nine-class panels identified panels that could locally distinguish all eight rates. Among these panels, class choice changed the allowable measurement error by approximately 18-fold for the same target uncertainty in inferred rates. A separate simulation within the fixed network showed that estimating rate-change combinations with weaker composition responses reduced bias but amplified measurement noise. We applied the framework to plasma data from 40 controls and 70 patients across 14 inherited glycosylation-disorder groups. A prespecified composition score for the expected pathway disturbance ranked among the top two in 13 of 14 groups. Network responses linked observed composition shifts to biosynthetic reaction classes. Expanding site models from five to eight reaction classes improved control-composition fit but reduced the number of sites at which all modelled rates were locally distinguishable and increased noise sensitivity. Rate inference thus requires sufficient panel size, separable rate responses, adequate measurement precision and a model that fits the composition. These requirements guide measurement design and help determine whether compositions support pathway-level interpretation or individual-rate inference. Author summary: Mass spectrometry measures the relative amounts of glycan forms attached to a protein site. We asked when these proportions can distinguish changes in the biosynthetic reactions that produce them. Because proportions sum to 100%, a panel contains one fewer independent quantity than its number of glycoform classes. Our eight-rate model therefore required at least nine classes. Which classes were measured strongly affected the measurement precision needed to distinguish the rates. We applied the framework to plasma data from inherited glycosylation disorders. The expected composition change ranked among the two strongest signals in 13 of 14 disorder groups. Network responses linked these changes to reaction groups. Most site models could distinguish small rate changes near a fitted control reference, but this did not establish recovery of patients’ true rates. A more detailed model reproduced compositions better while making individual rates harder to distinguish. The framework helps researchers select informative glycoforms, calculate measurement-precision requirements and assess which biosynthetic changes the data can support. Its conclusions depend on the assumed network, reference rates and measurement-error model.

  • Open Access

    GlycoMeSH: linking glycan structures to biomedical context for systematic enrichment analysis

    by Akihiro Kitani, Bingyuan Zhang, Koichi Himori, Yusuke Matsui

    Glycan identification has advanced, but glycan structures remain difficult to translate into reproducible biomedical context because reusable glycan-level annotations are sparse. We present GlycoMeSH, a resource that links glycans to Medical Subject Headings (MeSH) through an inference model, a traceable association database and a glycan-set enrichment workflow. GlycoMeSH-BERT recovered ~60% of literature-derived associations at recall@30 and expanded open-vocabulary MeSH coverage beyond closed-label baselines, without higher per-prediction accuracy. At matched candidate counts, its predictions showed motif-level semantic agreement comparable to those baselines, independently of the training labels. GlycoMeSH-DB contains 789,627 associations between 26,954 glycans and 20,302 MeSH terms. GlycoMeSH-EA returned enriched MeSH terms for glycan sets from glycomics and glycoproteomics datasets. Each association represents a biomedical context rather than a validated mechanism, and retains its source PMID or prediction score for audit. GlycoMeSH supplies the missing, evidence-traceable annotation layer that makes glycan sets directly analyzable by enrichment across glycoscience datasets.

  • Open Access

    GlycanGT: a pretrained graph transformer framework for glycan graph representation and generative learning

    by Akihiro Kitani, Bingyuan Zhang, Koichi Himori, Yusuke Matsui

    Motivation: Glycans are highly diverse biological sequences, but their functional understanding has lagged behind proteins and nucleic acids. Many glycans remain ambiguously annotated, limiting computational analyses. Existing computational approaches are primarily graph-based, capturing local structural features but struggling to model global patterns and incomplete sequences. Results: We present GlycanGT, a graph-transformer-based pretrained model for glycans. Glycans were represented as graphs of monosaccharides and glycosidic bonds, and the model was pretrained using a masked language modeling objective. GlycanGT demonstrated higher performance than existing methods across 8 benchmark tasks (e.g., 0.844 AUPRC for immunogenicity classification), and its embeddings formed biologically relevant clusters that recovered known N- and O-glycan categories. Moreover, GlycanGT accurately proposed candidates for ambiguous sequences, maintaining >80% top-5 accuracy for both monosaccharide and glycosidic bond predictions under high masking levels. Availability and implementation: The source code used in this study is available at https://github.com/matsui-lab/GlycanGT and archived on Zenodo (DOI: 10.5281/zenodo.18636040); pretrained model weights are provided via Hugging Face (https://huggingface.co/Akikitani295/GlycanGT). Contact: matsui.yusuke.d4@f.mail.nagoya-u.ac.jp Supplementary information: Supplementary data are available at Bioinformatics online.


  • Open Access

    GlycanGT: A Foundation Model for Glycan Graphs with Pretrained Representation and Generative Learning

    by Akihiro Kitani, Bingyuan Zhang, Koichi Himori, Yusuke Matsui

    Motivation: Glycans are highly diverse biological sequences, but their functional understanding has lagged behind that of proteins and nucleic acids. Many glycans remain incompletely characterized or ambiguously annotated, limiting computational analyses. Existing computational approaches are primarily graph-based, capturing local structural features but struggling to model global patterns and incomplete sequences. Results: We present GlycanGT, a foundation model for glycans built on a graph transformer architecture. Glycans were represented as graphs with monosaccharides as nodes and glycosidic bonds as edges, and the model was pretrained using a masked language modeling objective. GlycanGT demonstrated higher performance than existing methods across 8 benchmark classification tasks (e.g., 0.734 Macro-F1 in domain prediction and 0.844 AUPRC for immunogenicity classification), and its embeddings formed biologically meaningful clusters that recovered known N- and O-glycan categories. Moreover, GlycanGT accurately proposed candidates for ambiguous sequences, maintaining >80% top-5 accuracy for both monosaccharide and glycosidic bond predictions under high masking levels. Availability and implementation: The pretrained GlycanGT model weights and usage scripts are available on Hugging Face: https://huggingface.co/Akikitani295/GlycanGT . Additional scripts used for analyses in the paper are publicly available on GitHub: https://github.com/matsui-lab/GlycanGT . Contact: matsui.yusuke.d4@f.mail.nagoya-u.ac.jp


  • Open Access

    Mechanistic Insights into Cartilage Protection and Extracellular Matrix Remodeling: Transcriptome Analysis of Diclofenac Etalhyaluronate-Treated Knee Cartilage in Collagen-Induced Arthritis Rats and Cytokine-Stimulated Human Chondrocytes

    by Akihiro Kitani, Shuhei Takada, Shohei Nozaki, Kazuhiro Kojima, Kei Toyama, Takafumi Hashimoto, Jun Takeuchi, Akihiko Kawamoto, Takehito Fujita, Keiji Yoshioka

    Background: Diclofenac etalhyaluronate (DF-HA, SI-613/ONO-5704) is a conjugate of hyaluronic acid (HA) and diclofenac (DF), and its intra-articular injection is widely used for the treatment of osteoarthritis in Japan. While novel mechanisms of cartilage protection by DF-HA have been identified, a comprehensive analysis of the biological responses unique to DF-HA has not yet been conducted. Design: We used an RNA sequencing (RNA-seq) method to comprehensively analyze gene expression in the knee joint cartilage of arthritic rats and cytokine-stimulated chondrocytes. For the mechanistic analysis of DF-HA, genes that were downregulated or upregulated by DF-HA, HA, or DF were extracted. Pathway analysis was then performed on genes that specifically varied with DF-HA treatment. Results: In the cartilage of rats with collagen-induced arthritis, treatment with DF-HA, but not DF or HA, suppressed the extracellular matrix (ECM) remodeling pathway and promoted the parathyroid hormone/parathyroid hormone-related peptide receptor-mediated pathway, which regulates chondrocyte differentiation and bone/cartilage development. In cytokine-stimulated chondrocytes, DF-HA similarly suppressed the ECM remodeling pathway; specifically, gene expression changes in IGFBP4, MMP10, MMP13 , and TIMP1 were consistent with those observed in vivo . Conclusion: RNA-seq analysis of cartilage in arthritic rats and cytokine-stimulated chondrocytes provided molecular mechanistic insights, indicating that DF-HA treatment induced cartilage protection through the suppression of ECM remodeling.


  • Open Access

    Candidate Molecular Subtypes of Cognitive Resilience in Alzheimer’s Disease: A Multi-Cohort Machine Learning and Neuroimaging Study

    by Akihiro Kitani, Yusuke Matsui

    Background: Cognitive resilience (CR) in Alzheimer’s disease (AD) refers to preserved cognitive function despite substantial AD pathology. Diverse biological processes have been implicated in CR, including synaptic maintenance, neuroimmune regulation, and metabolic homeostasis. However, how these mechanisms are organized into molecularly distinct CR subtypes and relate to clinical and neuroanatomical heterogeneity remains unclear. Here, we applied a machine learning framework to multi-cohort transcriptomic, proteomic, and neuroimaging data to investigate molecular subtypes of CR in AD. Methods: RNA-seq data from the Religious Orders Study and Memory and Aging Project (ROSMAP) cohort were used to train machine learning models classifying individuals with AD pathology as CR or non-CR based on residual-based resilience scores. Model development and performance estimation used nested cross-validation to minimize information leakage. Final ROSMAP-trained models were evaluated in the independent Mount Sinai Brain Bank (MSBB) cohort. Model-derived genes were used for biological interpretation and hierarchical clustering of CR individuals. The subtype structure was further evaluated in the Alzheimer’s Disease Neuroimaging Initiative (ADNI) cohort using cerebrospinal fluid proteomics, MRI-derived brain measures, and longitudinal MMSE data. Results: Machine learning models showed modest but consistent predictive performance in ROSMAP, with out-of-fold AUROC values of 0.644–0.688. In the independent MSBB full cohort, AUROC values were 0.586–0.659, with improved discrimination in a top/bottom quartile analysis. Hierarchical clustering identified two major molecular subgroups among CR individuals in ROSMAP/MSBB RNA-seq data. A reduced 22-gene/protein signature showed a partial, cluster-like resemblance to this structure in ADNI cerebrospinal fluid proteomics. In ADNI, both projected CR subtypes showed preserved brain tissue-volume profiles and slower longitudinal MMSE decline compared with non-CR participants, whereas clear differences between CR subtypes were not observed. Differential CSF proteomic analysis suggested partially distinct molecular characteristics. Conclusions: These findings suggest that CR in AD may encompass molecularly heterogeneous, subtype-like profiles that converge on broadly preserved brain structure and slower cognitive decline. Our results provide a candidate framework for stratifying resilience-associated molecular phenotypes in AD and warrant prospective and experimental validation. We also developed the Resilience Gene Analyzer, a web-based platform for visualizing gene-level contributions to CR prediction ( https://igcore.cloud/GerOmics/REsilienceGeneAnalyzer/ ).

  • Open Access

    Integrative Network Analysis Reveals Novel Moderators of Aβ-Tau Interaction in Alzheimer’s Disease

    by Akihiro Kitani, Yusuke Matsui

    Background: Although interactions between amyloid-beta and tau proteins have been implicated in Alzheimer’s disease (AD), the precise mechanisms by which these interactions contribute to disease progression are not yet fully understood. Moreover, despite the growing application of deep learning in various biomedical fields, its application in integrating networks to analyze disease mechanisms in AD research remains limited. In this study, we employed BIONIC, a deep learning-based network integration method, to integrate proteomics and protein–protein interaction data, with an aim to uncover factors that moderate the effects of the Aβ-tau interaction on mild cognitive impairment (MCI) and early-stage AD. Methods: Proteomic data from the ROSMAP cohort were integrated with protein–protein interaction (PPI) data using a Deep Learning-based model. Linear regression analysis was applied to histopathological and gene expression data, and mutual information was used to detect moderating factors. Statistical significance was determined using the Benjamini-Hochberg correction (p < 0.05). Results: Our results suggested that astrocytes and GPNMB+ microglia moderate the Aβ-tau interaction. Based on linear regression with histopathological and gene expression data, GFAP and IBA1 levels and GPNMB gene expression positively contributed to the interaction of tau with Aβ in non-dementia cases, replicating the results of the network analysis. Conclusions: These findings indicate that GPNMB+ microglia moderate the Aβ-tau interaction in early AD and therefore are a novel therapeutic target. To facilitate further research, we have made the integrated network available as a visualization tool for the scientific community (URL: https://igcore.cloud/GerOmics/AlzPPMap ).