Citations

Clear

1526 citation(s) found for "Sun, Yu".

ATP1A3 as a target for isolating neuron‐specific extracellular vesicles from human brain and biofluids
Zhang, Zhengrong , Sultana, Nadia , Martens, Yuka A. , Sun, Min , Kanekiyo, Takahisa , Ikezu, Seiko , You, Yang , Sultana, Nadia, Ericsson, Maria , Ikezu, Tsuneya , Shaffer, Scott A.

Mayo Clinic

Alzheimer's & Dementia | January 9, 2025 | Volume: 20 | Issue: S2 | Pages: No page available

10.1002/alz.088090 Get It

AbstractBackgroundNeuron‐derived extracellular vesicles (NDEVs) are a valuable resource for understanding brain conditions and discovering neurodegenerative diseases biomarkers, notably Alzheimer’s disease (AD). Recent interest focuses on capturing neuron‐specific EVs from patient‐derived samples, characterizing their contents as a pathological reflection of the central nervous system (CNS). Our recent study identified ATPase Na+/K+ Transporting Subunit Alpha 3 (ATP1A3) as a prevalent neuron‐specific EV marker specifically expressed in brains. This study systematically analyzes neuronal EVs to assess the specificity of ATP1A3 to NDEVs and its potential as a target for NDEV pulldown from accessible biofluids for disease monitoring. Additionally, ATP1A3 is compared with other proposed NDEV markers.MethodWe applied immunoelectron microscopy to detect ATP1A3, L1 cell adhesion molecule (L1CAM) and neural cell adhesion molecule 1 (NCAM1) in EVs isolated from iPSC‐derived excitatory neurons, brain tissue, cerebrospinal fluid (CSF) and plasma. Neuronal EV enrichment was achieved through immunoaffinity isolation, utilizing anti‐ATP1A3, L1CAM and NCAM1 antibodies separately. Evaluation involved quantitative mass‐spectrometry, immunoblotting and ELISA. To compare the enrichment of neuronal markers, EVs isolated from CSF and plasma samples were analyzed using ExoView and Nanoimager. Additionally, we investigated the potential of ATP1A3+ plasma EVs as AD biomarker by comparing the value of amyloid‐beta peptide (Aβ) in ATP1A3+ plasma EVs measured by Nanoimager and other plasma AD biomarkers as determined by SIMOA.ResultATP1A3, identified as a neuron‐specific protein, exhibits substantial enrichment in NDEVs isolated from induced human neurons, brain tissue, CSF, and plasma samples, surpassing NCAM1 or L1CAM. Both single‐ and bulk‐EV analysis consistently demonstrate a higher enrichment of ATP1A3 associated EVs in human samples. Immunoprecipitation of ATP1A3+ EVs from human brain tissues reveals superior neuronal cell‐type specificity over NCAM1+ and L1CAM+ EVs by label‐free mass spectrometry. Increased Aβ levels are observed in ATP1A3+ EVs derived from CSF and plasma of AD cases by SIMOA. Moreover, Aβ+ populations in ATP1A3+ EVs from plasma can distinguish AD from mild cognitive impairment and control cases via Nanoimager compared to the conventional quantification of AD markers.ConclusionOur findings demonstrate that ATP1A3 as a promising target for isolating NDEVs from biofluids, offering potential diagnostic advancements in neurological research.
Ethical Use of Cadaveric Images in Anatomical Textbooks, Atlases, and Journals: A Consensus Response From Authors and Editors
Kim, Hee‐Jin , Logan, Bari M., Hutchings, R. T., Anand, Mahindra , Burns, Danny, Singh, Vishram, Peris‐Celda, Maria , Martı́nez-Soriano, Francisco, Apaydın, Nihal, Hanna, Amgad S. , Yoshioka, Nobutaka , Fernandez‐Miranda, Juan C. , Saremi, Farhood , Spratt, Jonathan D., Ho, Siew Yen , Mori, Shumpei , Komune, Noritaka , Watanabe, Koichi , Abrahams, Peter, Ibaragi, Soichiro , Loukas, Marios , Tubbs, R. Shane, Kim, Hee-Jin , Hur, Mi-Sun , Reina, Francisco , Tabira, Yoko , Carrera, Anna , Komune, Noritaka , Watanabe, Koichi , Reina, Miguel Angel , Loukas, Marios , Iwanaga, Joe , Akita, Keiichi , Ottone, Nicolás Ernesto , Nonaka, Yoichi , Shoja, Mohammadali M. , Tabira, Yoko , Prats‐Galino, Alberto , Andrés, José De , Reina, Miguel Ángel , Anderson, Robert Henry

Mayo Clinic

Clinical Anatomy | January 4, 2025 | Volume: 38 | Issue: 2 | Pages: 222-225

10.1002/ca.24258 | PMID 39754475 Get It

ABSTRACT Nowadays, consent to use donor bodies for medical education and research is obtained from the body donors and their families before the donation. Recently, the International Federation of Associations of Anatomists (IFAA) published guidelines that could restrict the appearance of cadaveric images in commercial anatomical resources such as textbooks and other educational products. These guidelines state that the donor must expressly consent to using such images for this purpose. Cadaveric photos and drawings made from dissections of cadavers have been used in anatomy textbooks and atlases for hundreds of years. They are invaluable for anatomy students and clinical/surgical practitioners. The IFAA guidelines should not restrict the use of those older books; to do so would infringe the rights of those seeking knowledge from these resources. As the images in such textbooks and atlases are anonymized and are used for teaching and research, and the donors and their families are informed about this before the donation, we believe no additional consent is needed. It is impossible to separate educational from “commercial” usage entirely in any situation, e.g., publications from publishers and the use of cadavers in medical schools. Therefore, our best efforts to avoid unethical use of cadaveric images by following traditional consent processes are still needed so that more people will reap the benefits from them. As senior textbook/atlas authors/editors from over 10 countries, we believe that using cadaveric images in anatomy textbooks is appropriate, and no additional consent should be necessary. Such usage falls within the good faith of professionals using these invaluable gifts.
Investigating the effect of Area Deprivation Index on the association of APOE‐ε4 and APOE‐ε2 with Alzheimer’s disease in a Multiethnic Autopsy Cohort
Yepez‐Sisalema, Geovanna, Pradeep, Aishwarya , Gonzalez‐Soto, Rous M, Nguyen, Thuy , Sotelo, Katie D., Reddy, Joseph S. , Allen, Mariet , Beach, Thomas G., Ertekin‐Taner, Nilüfer , Carrasquillo, Minerva M. , Yepez‐Sisalema, Geovanna, Kuchenbecker, Lindsey A. , Heckman, Michael G. , Serrano, Geidy E. , Prokop, Stefan , Dickson, Dennis W. , Carrasquillo, Minerva M.

Mayo Clinic

Alzheimer's & Dementia | January 3, 2025 | Volume: 20 | Issue: S1 | Pages: No page available

10.1002/alz.092762 Get It

AbstractBackgroundAfrican Americans (AA) and Latin Americans (LA) are at a higher risk of developing AD compared to non‐Hispanic whites (NHW) but are traditionally underrepresented in AD research. The disproportionate risk is likely multifactorial including differences in co‐morbidities and structural and social determinants of health (SSDoH). AD risk is thought to result from multiple genetic and environmental factors, and their interactions (GxE). However, the molecular mechanisms underlying these interactions are largely unknown. The best‐established genetic risk factor for AD is APOE‐ε4, while APOE‐ε2 is protective. Environmental factors, such as living in neighborhoods with greater socioeconomic disadvantage, have been associated with greater AD neuropathology and worse cognitive function. Studying the potential modifying role of neighborhood disadvantage regarding associations of APOE‐ε4 and APOE‐ε2 with AD risk using a socioeconomic metric like the Area Deprivation Index (ADI) in a multiethnic post‐mortem cohort could offer valuable insights for precision medicine.MethodWe analyzed a post‐mortem cohort of 421 multi‐ethnic donors (NHW: N = 234, AD = 70, non‐AD = 164; LA: N = 145, AD = 102, non‐AD = 43; AA: N = 42, AD = 24, non‐AD = 18), from Mayo Clinic Florida, Banner Sun Health Research Institute, and the University of Florida Brain Banks as part of the AMP AD Diversity initiative. This study used the same neuropathological criteria to categorize AD vs. non‐AD as in the MayoRNAseq study (syn5550404). ADI from the Neighborhood Atlas® was obtained from de‐identified addresses. Associations of APOE‐ε4, APOE‐ε2, and ADI with AD risk were examined using logistic regression models adjusted for age‐at‐death, sex, and ethnic group; interactions of APOE genotype with ADI were also assessed.ResultAPOE‐ε4 and APOE‐ε2 showed a significant association with AD diagnosis (APOE‐ε4combined cohort OR = 3.23, P = 3.14E‐08; APOE‐ε2combined cohort OR = 0.30, P = 0.003). However, contrary to prior findings, our study found no association between ADI and AD diagnosis [OR = 1.023 (per each 10 unit increase in ADI), P = 0.629]. No significant interactions of ADI with APOE‐ε4 or APOE‐ε2 regarding association with AD diagnosis were identified (P = 0.083, P = 0.781).ConclusionThis is the first study evaluating the potential interaction of neighborhood disadvantage with APOE on AD risk in a multi‐ethnic autopsy cohort. However, these findings require further investigation in larger, more well‐powered cohorts for validation.
Multi‐omics profiling of Alzheimer’s disease in multi‐ethnic populations
Reddy, Joseph S. , Allen, Mariet , Poehlman, William L. , Wang, Yanling , Johnson, Ryan , Ho, Charlotte C.G. , Linden, Abby Vander , Quicksall, Zachary , Chen, Xianfeng , Baheti, Saurabh , Nguyen, Thuy , Yepez‐Sisalema, Geovanna, Mitchell, Adriana O., Oatman, Stephanie R., Wang, Xue , Carrasquillo, Minerva M. , Beach, Thomas G., Golde, Todd E. , Haroutunian, Vahram , Jager, Philip L. De , Bennett, David A. , Seyfried, Nicholas T. , Ertekin‐Taner, Nilüfer , Initiative, Accelerating Medicines Partnership for Alzheimer’s Disease (AMP‐AD) Diversity, Heath, Laura , Johnson, Ryan, Wang, Erming , Ma, Yiyi , Lopes, Kátia de Paiva , Wang, Xue, Carrasquillo, Minerva M. , Runnels, Alexi , Teich, Andrew F. , Beach, Thomas G. , Serrano, Geidy E. , Barnes, Lisa L. , Wang, Minghui , Zhang, Bin , Greenwood, Anna, Initiative, Accelerating Medicines Partnership for Alzheimer’s Disease (AMP‐AD) Diversity

Mayo Clinic

Alzheimer's & Dementia | January 3, 2025 | Volume: 20 | Issue: S1 | Pages: No page available

10.1002/alz.090590 Get It

AbstractBackgroundAlzheimer’s disease (AD) is a devastating neurodegenerative disorder with few therapies to treat, mitigate or prevent its onset. Understanding of this disease is predominantly based on research in non‐Hispanic Whites (NHW) although AD disproportionately affects African Americans (AA) and Latin Americans (LA), underrepresented in AD research. To address this knowledge gap, the Accelerating Medicine Partnership for Alzheimer’s Disease (AMP‐AD) Diversity Working Group was launched to generate multi‐omics data from post‐mortem brain tissue from donors of predominantly AA and LA descent. Here, we focus on genomic and transcriptomics measures, made available through the AD Knowledge Portal (ADKP).MethodBrain tissue from dorsolateral prefrontal cortex (DLPFC), caudate nucleus (CN), superior temporal gyrus (STG) and temporal pole (TP) regions of AA (N = 299), LA (N = 344) and NHW (N = 532) donors was obtained through Mayo Clinic, Mt. Sinai School of Medicine, Emory, Columbia, Rush Alzheimer’s Disease Center, Banner Sun Health, and 1Florida. DNA extracted from DLPFC tissue samples (n = 641) was utilized for whole genome sequencing (WGS) at New York Genome Center (NYGC). RNAseq of 2,224 samples was conducted either at Mayo, Rush or NYGC. Sequencing reads were processed through GenomeGPS and MAPRseq pipelines for WGS and RNAseq data, respectively. Following quality control (QC), RNAseq data was normalized and assessed for sources of variation (SOV). Integration of WGS and RNAseq data to perform comparative eQTL analysis across multi‐ethnic populations and brain regions is ongoing.ResultWGS data for 621 donors, and RNAseq data generated at Mayo (n = 1,136 samples), Rush (n = 627), and NYGC (n = 779) sequencing centers, were deposited to the ADKP. Consensus processing of RNAseq data revealed site, sex, race/ethnicity, RIN, AD diagnosis, age at death and/or library batch as SOV across various sites and brain regions. Principal component analysis of gene expression measures revealed no major stratification by disease or brain region. Accompanying clinical metadata of donors have been distributed through ADKP.ConclusionThese multi‐omics data made available to the research community is expected to be an initial step towards bridging our data and knowledge gap to elucidate conserved and distinct AD related pathways across these underrepresented at‐risk populations.
Distribution of cognitive domains and neuropathology of cases with mild cognitive impairment
Adler, Charles, Mehta, Shyamal H. , Shill, Holly A. , Driver‐Dunckley, Erika, Belden, Christine M. , Beach, Thomas G., Ho, Andrew, Choudhury, Parichita , Shprecher, David , Driver‐Dunckley, Erika , Serrano, Geidy E. , Atri, Alireza , Tremblay, Cécilia

Mayo Clinic

Alzheimer's & Dementia | January 3, 2025 | Volume: 20 | Issue: S3 | Pages: No page available

10.1002/alz.092655 Get It

AbstractBackgroundMild Cognitive Impairment (MCI) is a pre‐dementia state where impaired cognitive domains (memory, executive, visuospatial processing, language) may predict underlying pathology, e.g. Alzheimer’s disease(AD), Lewy body (LB), and other. Delineating neuropathological heterogeneity in MCI and their associations with neuropsychological profiles are important in identifying individuals at risk of further deterioration. We aimed to characterize the neuropathological heterogeneity in cases with a diagnosis of MCI.MethodCases with a final consensus cognitive diagnosis of MCI (n = 128) prior to death were selected from the Arizona Study of Aging and Neurodegenerative disorders (AZSAND). Groups were divided into amnestic (aMCI) and non‐amnestic MCI (naMCI), single and multi‐domain impairment, and specific domain impairment (executive, language, visuospatial) was explored. Presence and severity of neuropathology including AD, LB, cerebral white matter rarefaction (CWMR), cerebral amyloid angiopathy (CAA), cerebral infarcts and microinfarcts (cortical and subcortical), TAR DNA‐binding protein 43 (TDP‐43), and apolipoprotein E (ApoE) status were compared between MCI subtypes.ResultThe number of cases meeting neuropathological criteria for AD was not different between aMCI (36.4%) and naMCI (22.6%) while co‐existent PD was less frequent in aMCI (19.5% vs 38.7%). When compared to naMCI (n = 31), the aMCI (n = 85) group had an older age of death (p = 0.001), less educational years (p = 0.04), a higher number of subcortical infarcts (p = 0.012), a higher regional density of neurofibrillary tangles (p = 0.027) and a lower density of LB (p = 0.015) in the temporal lobe, and CWMR in tended to be more severe in the frontal lobe (p = 0.052). Those with single domain aMCI had less CWMR (p = 0.001) compared to the multidomain. Within the naMCI group, those with single domain MCI had higher proportions of CAA (p = 0.042), and lower LB density (p = 0.019) compared to multidomain. CWMR was more severe (p = 0.026) inConclusionA high heterogeneity was observed between MCI subtypes. While AD pathology was similar between groups, regional anatomical differences were observed. Vascular pathology including subcortical infarcts, CWMR and CAA appear to be commonly associated with both aMCI and naMCI respectively while CWMR was associated with executive dysfunctions. CWMR, CAA and LBs appear to be more frequent in multidomain aMCI and naMCI respectively.
Development of a composite score to predict Lewy body pathology burden
Zhang, Nan , Belden, Christine M. , Adler, Charles, Shill, Holly A. , Mehta, Shyamal H. , Driver‐Dunckley, Erika, Beach, Thomas G. , Reiman, Eric M., Choudhury, Parichita , Chen, Kewei , Tremblay, Cécilia , Ho, Andrew, Driver‐Dunckley, Erika , Shprecher, David , Serrano, Geidy E. , Reiman, Eric M. , Atri, Alireza

Mayo Clinic

Alzheimer's & Dementia | January 3, 2025 | Volume: 20 | Issue: S3 | Pages: No page available

10.1002/alz.092056 Get It

AbstractBackgroundLewy body (LB) diseases can present with overlapping prodromal, cognitive, motor, autonomic or neuropsychiatric symptoms. Intuitively, greater symptom severity should correlate with greater pathological burden, but this has not been consistently shown. LB pathology does not translate to clinical expression in Incidental LB disease. While several composite scores/toolkits utilize overlapping schemes for diagnosis, a composite data‐derived score to predict LB pathologic burden is lacking. We aimed to utilize an autopsy‐confirmed cohort, and validated, brief, and standardized clinical, functional, pathologic assessments, and machine learning (ML)/quantitative modeling techniques to identify clusters and importance rankings among assessments/measures to predict LB pathologic severity and density.MethodA total of 234 subjects with neuropathological finding of LB at autopsy in the Arizona Study of Aging and Neurodegenerative Disorders (AZSAND) who were classified as Cognitively Unimpaired or Mild Cognitive Impairment after their first study clinical evaluation were included. Of these, 145 subjects had complete scores for 32 of 46 measures (clinical, cognitive, behavioral and functional) that were utilized as predictors of LB pathology. Olfactory function was assessed using the University of Pennsylvania Smell Identification Test (UPSIT). LB severity was assessed by the Unified Staging System for Lewy Body Disorders (USSLB). Several ML algorithms and quantitative prediction models were explored and compared: artificial neural network (ANN), partial least square regression (PLSR), support vector regression (SVR), relevance vector regression (RVR) and ensemble forest regression (EFR) with leave‐one‐out (LOO) scheme. Game‐theory based Shapley methods assessed the impact, including rankings, consistency and magnitude, of model predictors.ResultRVR predicted aggregate LB density with large effect size (R2 = 0.691, p<0.00001). ANN predicted USSLB severity stage with large effect size (R2 = 0.74, p<2.5e‐43). All other ML algorithms/models provided substantial prediction. Across all models, UPSIT was the most influential predictor (>65%), followed by Controlled Oral Word Association Test (COWAT) and age.ConclusionThese preliminary and exploratory results support the utilization of ML techniques/models to assess LB pathologic burden with key measures collected in relatively small samples. UPSIT was consistently ranked highest impactful among clinical and functional measures/predictors. Integrated together in ML/data‐derived composites, UPSIT, COWAT and other clinical characteristics may be of antemortem utility to predict USSLB stages.
Optimal Population Screening Strategies for Liver Fibrosis Associated With Metabolic Dysfunction-Associated Steatotic Liver Disease
Kim, W. Ray , Charu, Vivek , Chung, Nakia , Kwong, Allison J. , Kwo, Paul Y. , Török, Natalie J. , Asch, Steven M. , Kim, Sun H. , Mannalithara, Ajitha , Chung, Nakia , Kwong, Allison , Kwo, Paul Y , Torok, Natalie J , Asch, Steven M , Kim, Sun H

Mayo Clinic

American Journal of Gastroenterology | January 3, 2025 | Volume: 120 | Issue: 10 | Pages: 2312-2320

10.14309/ajg.0000000000003268 | PMID 39749918 Get It

INTRODUCTION: Metabolic dysfunction-associated steatotic liver disease (MASLD) is an important public health threat, potentially leading to chronic liver disease and liver cancer. Current guidelines recommend using the Fibrosis-4 score for initial identification of subjects at risk of future complications. We formulate a novel population screening strategy based on the Steatosis-Associated Fibrosis Estimator (SAFE) score, recently developed for MASLD risk stratification in primary care. METHODS: We interrogated the National Health and Nutrition Examination Survey data, 2017–20, in which a sample of subjects representative of US civilian population underwent vibration-controlled transient elastography (VCTE). The current guideline and a new, SAFE-based proposal were applied to these data to project the number of subjects to be diagnosed with liver fibrosis gauged by liver stiffness measurement (LSM), including significant (LSM ≥8 kPa) and advanced (LSM ≥12 kPa) fibrosis, as well as the number of VCTEs to be performed. RESULTS: In the survey data, 2,691 subjects, projecting to 75.8 million US adults, were found to have MASLD, of whom 11% had LSM 8–12 kPa and 6% LSM ≥12 kPa. When the current guideline was applied, 18.1 million VCTEs would be needed to diagnose 3.5 million subjects with LSM ≥8 kPa and 1.7 million subjects with LSM ≥12 kPa. In comparison, a new approach based on the SAFE score would detect 4.9 million with LSM ≥8 kPa and 2.5 million subjects with LSM ≥12 kPa (37% and 45% improvement over the current guideline, respectively), while requiring 5.0 million fewer VCTEs (28% reduction). DISCUSSION: The proposed population risk stratification approach using the SAFE score is simpler and substantially more effective, yielding more subjects with liver fibrosis while requiring less resources compared with the currently recommended algorithm.
Consensus Recommendations for Studies of Outflow Facility and Intraocular Pressure Regulation Using Ex Vivo Perfusion Approaches
Acott, Ted S. , Fautsch, Michael P. , Mao, Weiming , Huang, Alex S. , Kelley, Mary J. , Aga, Mini , Borrás, Terete, Bovenkamp, Diane E. , Chowdhury, Uttio Roy, Clark, Abbot F. , Dibas, Mohammed, Elliott, Michael H. , Faralli, Jennifer A., Gong, Haiyan , Kaufman, Paul L., Keller, Kate E. , Kuehn, Markus H. , Li, Hoi Lam, Lin, Shan C. , Liu, Yutao , McDonnell, Fiona , McDowell, Colleen M. , Mzyk, Philip , Nair, K. Saidas , Overby, Darryl R. , Peters, Donna M. , Raghunathan, VijayKrishna, Rao, Ponugoti Vasantha , Roddy, Gavin W. , Sharif, Najam A. , Shim, Myoung Sup , Sun, Yang , Thomson, Benjamin R. , Willoughby, Colin E. , Freddo, Thomas F., Fuchshofer, Rudolf , Hill, Kamisha R. , Karimi, Alireza , Kizhatil, Krishnakumar , Kopcyznski, Casey C., Patel, Gaurang , Pattabiraman, Padmanabhan P. , Prasanna, Ganesh , Samples, E. Griffen, Samples, John R., Steel, Cynthia, Subramanian, Preeti, Sugali, Chenna Kesavulu , Sherwood, Joseph M. , Wong, Cydney A. , Zode, Gulab , White, Elizabeth , Liu, Yutao , Ethier, C. Ross , Bhattacharya, Sanjoy K. , Du, Yiqin , Herberg, Samuel , Johnstone, Murray , Kelly, Ruth A. , Križaj, David , Lieberman, Raquel L. , McLellan, Gillian J. , Raghunathan, VijayKrishna, Rao, Ponugoti Vasantha , Toris, Carol B. , Zhang, Hao F. , Liton, Paloma B. , Peng, Michael , Reina‐Torres, Ester , Strohmaier, Clemens , Youngblood, Hannah A. , Stamer, William Daniel

Mayo Clinic

Investigative Ophthalmology & Visual Science | December 18, 2024 | Volume: 65 | Issue: 14 | Pages: 32

10.1167/iovs.65.14.32 | PMID 39693082 Get It

A comprehensive benchmark study of methods for identifying significantly perturbed subnetworks in cancer
Yang, Le , Chen, Runpu , Goodison, Steve , Sun, Yijun

Mayo Clinic

Briefings in Bioinformatics | December 16, 2024 | Volume: 26 | Issue: 1 | Pages: No page available

10.1093/bib/bbae692 | PMID 39737568 Get It

Abstract Network-based methods utilize protein–protein interaction information to identify significantly perturbed subnetworks in cancer and to propose key molecular pathways. Numerous methods have been developed, but to date, a rigorous benchmark analysis to compare the performance of existing approaches is lacking. In this paper, we proposed a novel benchmarking framework using synthetic data and conducted a comprehensive analysis to investigate the ability of existing methods to detect target genes and subnetworks and to control false positives, and how they perform in the presence of topological biases at both gene and subnetwork levels. Our analysis revealed insights into algorithmic performance that were previously unattainable. Based on the results of the benchmark study, we presented a practical guide for users on how to select appropriate detection methods and protein–protein interaction networks for cancer pathway identification, and provided suggestions for future algorithm development.
A self-attention-driven deep learning framework for inference of transcriptional gene regulatory networks
Zhong, Le, Chen, Zhuobin, Yu, Yanjia , Yu, Dan, Qin, Jing , Wang, Junwen , Yu, Dan , Qin, Jing , Wang, Junwen , Liu, Yong , Yan, Bin, Chen, Zhuobin

Mayo Clinic

Briefings in Bioinformatics | December 16, 2024 | Volume: 26 | Issue: 1 | Pages: No page available

10.1093/bib/bbae639 | PMID 39679439 Get It

Abstract The interactions between transcription factors (TFs) and the target genes could provide a basis for constructing gene regulatory networks (GRNs) for mechanistic understanding of various biological complex processes. From gene expression data, particularly single-cell transcriptomic data containing rich cell-to-cell variations, it is highly desirable to infer TF–gene interactions (TGIs) using deep learning technologies. Numerous models or software including deep learning–based algorithms have been designed to identify transcriptional regulatory relationships between TFs and the downstream genes. However, these methods do not significantly improve predictions of TGIs due to some limitations regarding constructing underlying interactive structures linking regulatory components. In this study, we introduce a deep learning framework, DeepTGI, that encodes gene expression profiles from single-cell and/or bulk transcriptomic data and predicts TGIs with high accuracy. Our approach could fuse the features extracted from Auto-encoder with self-attention mechanism and other networks and could transform multihead attention modules to define representative features. By comparing it with other models or methods, DeepTGI exhibits its superiority to identify more potential TGIs and to reconstruct the GRNs and, therefore, could provide broader perspectives for discovery of more biological meaningful TGIs and for understanding transcriptional gene regulatory mechanisms.