The Chinese University of Hong Kong
  • Hong Kong, New Territories, China
Recent publications
Inference of population structure from genetic data plays an important role in population and medical genetics studies. With the advancement and decreasing cost of sequencing technology, the increasingly available whole genome sequencing data provide much richer information about the underlying population structure. The traditional method (Patterson, Price, and Reich, 2006) originally developed for array-based genotype data for computing and selecting top principal components that capture population structure may not perform well on sequencing data for two reasons. First, the number of genetic variants p is much larger than the sample size n in sequencing data such that the sample-to-marker ratio n/p is nearly zero, violating the assumption of the Tracy-Widom test used in their method. Second, their method might not be able to handle the linkage disequilibrium well in sequencing data. To resolve those two practical issues, we propose a new method called ERStruct to determine the number of top informative principal components based on sequencing data. More specifically, we propose to use the ratio of consecutive eigenvalues as a more robust test statistic, and then we approximate its null distribution using modern random matrix theory. Both simulation studies and applications to two public data sets from the HapMap 3 and the 1000 Genomes Projects demonstrate the empirical performance of our ERStruct method.
Over 300 billion of cells die every day in the human body, producing a large number of endogenous apoptotic extracellular vesicles (apoEVs). Also, allogenic stem cell transplantation, a commonly used therapeutic approach in current clinical practice, generates exogenous apoEVs. It is well known that phagocytic cells engulf and digest apoEVs to maintain the body's homeostasis. In this study, we show that a fraction of exogenous apoEVs is metabolized in the integumentary skin and hair follicles. Mechanistically, apoEVs activate the Wnt/β-catenin pathway to facilitate their metabolism in a wave-like pattern. The migration of apoEVs is enhanced by treadmill exercise and inhibited by tail suspension, which is associated with the mechanical force-regulated expression of DKK1 in circulation. Furthermore, we show that exogenous apoEVs promote wound healing and hair growth via activation of Wnt/β-catenin pathway in skin and hair follicle mesenchymal stem cells. This study reveals a previously unrecognized metabolic pathway of apoEVs and opens a new avenue for exploring apoEV-based therapy for skin and hair disorders.
Background Parkinson’s disease (PD) is a prevalent and debilitating condition. Conventional medications cannot control all symptoms and may inflict adverse effects. A survey reported that Chinese herbal medicine (CHM) is frequently sought. Existing CHM trials were contradictory and often of poor quality due to lack of methodological rigor. A national clinical guideline was drafted in China with diagnostic criteria and treatment strategy of Chinese medicine (CM) patterns subgroups of PD. The suggested CHM were found to exhibit neuroprotective effect in in vitro and in vivo studies. This trial aims to preliminarily assess the effect of CHM prescribed based on pattern differentiation on PD symptoms and patients’ quality of life, and evaluate the feasibility of the trial design for a future large-scale trial. Methods This trial will be a pilot assessor- and data analyst blind, add-on, randomised, controlled, pragmatic clinical trial. 160 PD patients will be recruited and randomised into treatment or control groups in a 1:1 ratio. The trial will be conducted over 32 weeks. PD patients in the treatment group will be stratified into subgroups based on CM pattern and receive CHM accordingly in addition to conventional medication (ConM). The control group will receive ConM only. The primary outcome will be part II of the Movement Disorder Society Sponsored Revision of Unified Parkinson’s Disease Rating Scale (MDS-UPDRS). Secondary outcomes will include part and total scores of MDS-UPDRS, domain and total scores of Non-motor symptom scale (NMSS). Adverse events will be monitored by monthly follow-ups and questionnaires. Mixed models will be used to analyse data by Jamovi and R. Expected outcomes The success of our trial will show that the pragmatic design with subgroup differentiation is feasible and can produce reliable results. It will also provide preliminary data of the effect of CHM on improving clinical outcomes and quality of PD patients. Data collected will be used to optimize study design of the future large-scale clinical study. Ethical clearance Ethical clearance of this study was given by the Research Ethics Committee of Hong Kong Baptist University (REC/20-21/0206). Trial registration This trial is registered on (NCT05001217, Date: 8/10/2021, ). Type of manuscript: clinical trial protocol (date: 3 rd November, 2021, version 1)
Single-stranded ends of double-stranded DNA (jagged ends) are more abundant in urinary DNA than in plasma DNA. However, the lengths of jagged ends in urinary DNA remained undetermined, as a previous method used for urinary DNA jagged end sequencing analysis (Jag-seq) relied on unmethylation at CpG sites, limiting the resolution. Here, we performed high-resolution Jag-seq analysis using methylation at non-CpG cytosine sites, allowing determination of exact length of jagged ends. The urinary DNA bore longer jagged ends (~26-nt) than plasma DNA (~17-nt). The jagged end length distribution displayed 10-nt periodicities in urinary DNA, which were much less observable in plasma DNA. Amplitude of the 10-nt periodicities increased in patients with renal cell carcinoma. Heparin treatment of urine diminished the 10-nt periodicities. The urinary DNA jagged ends often extended into nucleosomal cores, suggesting potential interactions with histones. This study has thus advanced our knowledge of jagged ends in urine DNA.
With rapid development of 5G communication technologies, electromagnetic interference (EMI) shielding for electronic devices has become an urgent demand in recent years, where the development of corresponding EMI shielding materials against detrimental electromagnetic radiation plays an essential role. Meanwhile, the EMI shielding materials with high flexibility and functional integrity are highly demanded for emerging shielding applications. Hitherto, a variety of flexible EMI shielding materials with lightweight and multifunctionalities have been developed. In this review, we not only introduce the recent development of flexible EMI shielding materials, but also elaborate the EMI shielding mechanisms and the index for "green EMI shielding" performance. In addition, the construction strategies for sophisticated multifunctionalities of flexible shielding materials are summarized. Finally, we propose several possible research directions for flexible EMI shielding materials in near future, which could be inspirational to the fast-growing next-generation flexible electronic devices with reliable and multipurpose protections as offered by EMI shielding materials.
Organic luminogens with room temperature phosphorescence (RTP) have been paid great attention and developed rapidly for their wide application values. Until now, the internal mechanism and source of phosphorescence are still obscure, especially for the relationship between molecular dimer and RTP emission. Hence, we designed and synthesized eight phenothiazine 5,5-dioxide derivatives to directly reveal how the monomer and dimer in packing affect the RTP behavior. Dimers with strong π-π stacking ( θ < 20.66°; d < 3.86 Å) lead to pure triplet excimer emission, while those with weak π-π stacking (27.02°< θ < 40.64°; 3.84 Å < d < 4.41 Å) contribute to dual RTP emissions of both monomer and triplet excimer. The valuable information of this work would promote the further development of this research field, as well as others in aggregate.
Objective To systematically review the clinical efficacy and safety of Chinese herbal medicine (CHM) with and without Western medicine (WM) for different severity of COVID-19. Methods CNKI, PubMed, Wanfang Database,, Embase, ChiCTR and ICTRP were searched from 01 Jan, 2020 to 30 Jun, 2021. Two authors independently assessed all the randomized clinical trials (RCTs) for trial inclusion, data extraction and quality assessment. Meta-analysis was conducted using Review Manager software (RevMan 5.4.1). Evidence was assessed using Grading of Recommendations Assessment, Development, and Evaluation (GRADE). Primary outcomes included total effectiveness rate. Secondary outcomes included improvements in symptom improvement and total adverse event rate. Different severity of COVID-19 patients was assessed in subgroup analysis. This study was registered with INPLASY, INPLASY202210072. Results 22 high quality RCTs involving 1789 participants were included. There were no trial used CHM alone nor compare placebo or no treatment. Compared with WM, combined CHM and WM (CHM-WM) treatment showed higher total effectiveness rate, lower symptom scores of fever, cough, fatigue, dry throat and pharyngalgia, shorter mean time to viral conversion, better Computerized Tomography (CT) image and blood results, fewer total adverse events and worse conditions (P < 0.05). Subgroup analysis showed that the total effectiveness rate of combined CHM-WM group was significantly higher than WM group, especially for mild and moderate patients. No significant differences in mortality and adverse events were found between combined CHM-WM and WM treatment. No serious adverse events and long-term outcomes were reported. Conclusion Current evidence supported the therapeutic effects and safety of combined CHM-WM treatment on COVID-19, especially for patients with mild and moderate symptoms. Long-term effects of therapy are worthy in further study.
The wake effect is the major obstacle to reaching the maximum power generation for wind farms, since choosing the suitable wake model that satisfies both computational cost and accuracy is a difficult task. Deep Reinforcement Learning (DRL) is a powerful data-driven method that can learn the optimal control policy without modeling the environment. However, the “trial and error” mechanism of DRL may cause high costs during the learning process. To address this issue, we propose an ensemble-based DRL wind farm control framework. Under this framework, a new algorithm called Actor Bagging Deep Deterministic Policy Gradient (AB-DDPG) is proposed, which combines the actor-network bagging method with the Deep Deterministic Policy Gradient. The gradient of the proposed method is proved to be consistent with the DDPG method. The experiment results in WFSim show that AB-DDPG can learn the optimal control policy with lower learning cost and a more robust learning process.
Background Macrophage phenotypes switch from proinflammatory (M1) to anti-inflammatory (M2) following myocardial injury. Implanted stem cells (e.g., induced pluripotent stem cells (iPSCs)) for cardiomyogenesis will inevitably contact the inflammatory environment at the myocardial infarction site. To understand how the macrophages affect the behavior of iPSCs, therefore, improve the therapeutic efficacy, we generated three macrophage subtypes and assessed their effects on the proliferation, cardiac differentiation, and maturation of iPSCs. Methods M0, M1, and M2 macrophages were polarized using cytokines, and their properties were confirmed by the expression of specific markers using reverse transcription-quantitative polymerase chain reaction (RT-qPCR) and immunofluorescence. The effects of macrophages on iPSCs were studied using Transwell co-culture models. The proliferative ability of iPSCs was investigated by cell counting and CCK-8 assays. The cardiac differentiation ability of iPSCs was determined by the cardiomyocyte (CM) yield. The maturation of CM was analyzed by the expression of cardiac-specific genes using RT-qPCR, the sarcomere organization using immunofluorescence, and the mitochondrial function using oxidative respiration analysis. Results The data showed that the co-culture of iPSCs with M0, M1, or M2 macrophages significantly decreased iPSCs’ proliferative ability. M2 macrophages did not affect the CM yield during the cardiac differentiation of iPSCs. Still, they promoted the maturation of CM by improving sarcomeric structures, increasing contractile- and ion transport-associated gene expression, and enhancing mitochondrial respiration. M0 macrophages did not significantly affect the cardiomyogenesis ability of iPSCs during co-culture. In contrast, co-culture with M1 macrophages significantly reduced the cardiac differentiation and maturation of iPSCs. Conclusions M1- or M2-polarized macrophages play critical roles in the proliferation, cardiac differentiation, and maturation of iPSCs, providing knowledge to improve the outcomes of stem cell regeneration therapy.
Type 1 diabetes (T1D) is an autoimmune disease that selectively destroys insulin-producing β-cells in the pancreas. An unmet need in diabetes management, current therapy is focussed on transplantation. While the reprogramming of progenitor cells into functional insulin-producing β-cells has also been proposed this remains controversial and poorly understood. The challenge is determining why default transcriptional suppression is refractory to exocrine reactivation. After the death of a 13-year-old girl with established insulin-dependent T1D, pancreatic cells were harvested in an effort to restore and understand exocrine competence. The pancreas showed classic silencing of β-cell progenitor genes with barely detectable insulin ( Ins ) transcript. GSK126, a highly selective inhibitor of EZH2 methyltransferase activity influenced H3K27me3 chromatin content and transcriptional control resulting in the expression of core β-cell markers and ductal progenitor genes. GSK126 also reinstated Ins gene expression despite absolute β-cell destruction. These studies show the refractory nature of chromatin characterises exocrine suppression influencing β-cell plasticity. Additional regeneration studies are warranted to determine if the approach of this n-of-1 study generalises to a broader T1D population.
In this paper, we consider a class of almost periodically forced harmonic oscillators x¨+τ2x=ϵf(t,x)\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\begin{aligned} \ddot{x}+\tau ^2 x=\epsilon f(t,x) \end{aligned}$$\end{document}where τ∈A\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\tau \in {{\mathcal {A}}}$$\end{document} with A\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${{\mathcal {A}}}$$\end{document} being a closed interval not containing zero, the forcing term f is real analytic almost periodic functions in t with the infinite frequency ω=(⋯,ωλ,⋯)λ∈Z\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\omega =(\cdots ,\omega _\lambda ,\cdots )_{\lambda \in {{\mathbb {Z}}}}$$\end{document}. Using the modified Kolmogorov–Arnold–Moser (or KAM Arnold (Uspehi Mat. Nauk 18(5 (113)):13–40, 1963), Moser (Nachr. Akad. Wiss. Göttingen Math.-Phys. Kl. II 1962:1–20 1962), Kolmogorov (Dokl. Akad. Nauk SSSR (N.S.) 98:527–530 1954)) theory about the lower dimensional tori, we show that there exists a positive Lebesgue measure set of τ\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\tau $$\end{document} contained in A\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${{\mathcal {A}}}$$\end{document} such that the harmonic oscillators has almost periodic solutions with the same frequencies as f. The result extends the earlier research results with the forcing term f being quasi-periodic.
Dual-comb spectroscopy (DCS) has been an emerging spectrometric technique with high resolution, high sensitivity, broad spectral range and fast acquisition speed. However, it is hard to achieve high long-term stability and high mutual coherence with low system complexity, which hinders the widespread usage in field-deployed applications. In this paper, we demonstrate a simplified free-running dual-comb spectrometer with long-term stability and mode-resolved resolution. Two free-running mode-locked fiber lasers are environment-shared to achieve effective common-mode noise suppression, which enables long-term aliasing-free dual-comb interferograms (IGMs). Post-proposing error correction algorithm is then utilized to recover full mutual coherence and realize coherent averaging. As a result, the transmittance spectra of H¹³CN and C2H2 gases with high resolution and high signal-to-noise ratio (SNR) can be measured. To the best knowledge of us, this is the first demonstration of coherent-averaged DCS based on environment-shared free-running fiber lasers, which is promising for promoting the dual-comb spectrometer out of the laboratory and realizing field-deployed applications.
We consider a multi-period optimal power flow (OPF) problem in transmission systems with a generic network topology subject to stochastic N−1 contingencies (disruptions). We propose a multi-stage stochastic convex programming formulation for such a disruption problem and solve it using an efficient decomposition algorithm. We compare the disruption optimization formulation with the conventional N−1 security-constrained OPF formulation and show that our proposed model can utilize the generation resource more efficiently in a realistic contingency setting. We also present computational results comparing multiple convex relaxation schemes of the non-convex alternating current OPF (ACOPF) problem in transmission networks under stochastic disruptions. Through this, we provide insights on the performance of the policy trained by convex relaxations of an ACOPF problem.
Price competition among electric vehicle (EV) charging stations is as fierce as the competition among gas stations. Nash equilibrium (NE) is a solution concept that can characterize a competition’s efficient and stable state. However, calculation of the equilibrium is often time-consuming and requires complete information on the charging stations. Rapidly changing charging stations often hinder reaching equilibrium. In this study, we analyze price competition with service capacity constraints and use an ordinal potential game framework to investigate the structure of the competition. By constructing the ordinal potential function, the equilibrium characterization is converted to identifying the solution through a single-objective optimization. We further propose a decentralized algorithm to enable effective price coordination to achieve equilibrium with maximized social welfare. To preserve the privacy of charging stations from internal collusion and external attacks, an advanced secure multi-party computation technology known as the Paillier Cryptosystem is customized for our proposed decentralized algorithm. Numerical studies based on field data suggest the significance of our framework.
The increasing penetration of distributed energy resources (DERs) brings the valuable flexibility to the system as well as challenges, which requires the coordination between transmission system operator (TSO) and distribution system operators (DSOs). This paper quantitatively models the goals and constraints of TSO and DSOs in procuring flexibility resources from DERs and proposes a flexibility market. We resolve the potential conflicts between TSO and DSOs in the flexibility market via mutual economic compensation, specified by Nash Bargaining (NB) theory. Two alternative TSO-DSOs interaction modes, i.e., the DSO-managed mode and the hybrid-managed mode, and two detailed payment schemes are discussed and compared. The hybrid-managed mode guarantees efficiency and fairness, while the DSO-managed mode is less complicated and can protect information privacy. Numerical studies further validate our findings.
Quantitative, sensitive and in situ detection of ammonia (NH3) is important for industrial process monitoring and emission characterization. We demonstrate high-temperature NH3 detection using heterodyne phase-sensitive dispersion spectroscopy (HPSDS) with a quantum cascade laser (QCL) emitting at 9.06 μm. The injection current of the QCL was modulated at 360 MHz to generate a three-tone laser beam and to maximize the dispersion signal of NH3 at the target temperature of 573 K and pressure of 100 Torr. The laser beam was directed into an optical probe, which can be inserted into the flue gas for in situ monitoring, to obtain an effective path length of 95 cm. The developed NH3 dispersion sensor shows an excellent linear response over the concentration range of 5–100 ppm (parts per million) and a minimum detection limit (MDL) of 0.69 ppm. The MDL of the current NH3 dispersion sensor can be further improved to 70 ppb (parts per billion) at a longer measurement time of 50 s. Compared to absorption spectroscopy, dispersion spectroscopy merits the advantage of immunity to laser power fluctuations caused by unwanted disturbance. The developed sensor is promising for NH3 slip monitoring in the selective catalytic reduction (SCR) used in power plants.
A dual-gas optical sensor was developed using frequency-division multiplexed (FDM) scanned-wavelength modulation spectroscopy for simultaneous detection of methane (CH4) and acetylene (C2H2). Two infrared lasers were combined to exploit C2H2 and CH4 absorption lines centered at 6539.45 cm⁻¹ and 6057.09 cm⁻¹, respectively. Scanned-WMS with the second-harmonic detection (WMS-2f) and a compact astigmatic multipass gas cell (36-m path length in a 0.3 L volume) were employed to improve detection sensitivity. Simultaneous dual-gas detection was realized by modulating each laser at a different frequency (5 kHz for 1529 nm and 13 kHz for 1650 nm) and software lock-in amplifiers were used to obtain the corresponding WMS-2f signals from a single photodetector. The Allan deviation analysis was performed to assess the long-term stability and minimum detection limit (MDL) of the sensor system. The MDL of 0.92 ppm for CH4 and 0.64 ppm for C2H2 was achieved at 0.02-s integration time, which can be further improved to 55 ppb for CH4 and 10 ppb for C2H2 at an optimum integration time of 10-s and 15-s, respectively. Such a dual-gas sensor has a promising potential for detecting dissolved gases (CH4 and C2H2) in transformer oil for fault diagnosis due to its high sensitivity, hardware simplicity, and fast temporal resolution.
In this paper, we analyze the landscape of the true loss of neural networks with one hidden layer and ReLU, leaky ReLU, or quadratic activation. In all three cases, we provide a complete classification of the critical points in the case where the target function is affine and one-dimensional. In particular, we show that there exist no local maxima and clarify the structure of saddle points. Moreover, we prove that non-global local minima can only be caused by ‘dead’ ReLU neurons. In particular, they do not appear in the case of leaky ReLU or quadratic activation. Our approach is of a combinatorial nature and builds on a careful analysis of the different types of hidden neurons that can occur.
Institution pages aggregate content on ResearchGate related to an institution. The members listed on this page have self-identified as being affiliated with this institution. Publications listed on this page were identified by our algorithms as relating to this institution. This page was not created or approved by the institution. If you represent an institution and have questions about these pages or wish to report inaccurate content, you can contact us here.
16,795 members
Benny Chow
  • Stanley Ho Center for Emerging Infectious Diseases
Dexing Zhang
  • The Jockey Club School of Public Health and Primary Care
Sien Lin
  • Department of Orthopaedics and Traumatology
Vincent H Lee
  • School of Pharmacy
KC Wong
  • Department of Orthopaedics and Traumatology
Hong Kong, New Territories, China