Juan Helen Zhou

Bio: Juan Helen Zhou is an academic researcher from Nanyang Technological University. The author has contributed to research in topics: Computer science & Medicine. The author has an hindex of 2, co-authored 2 publications receiving 192 citations.

An Ultra-Scalable Ensemble Clustering Method for Cell Type Recognition Based on scRNA-seq Data of Alzheimer's Disease

Abstract: Aiming at the clustering problem in single cells, considering the high dimensionality and sparsity of the data, we propose to apply an ultra-scalable ensemble clustering (U-SENC) algorithm to single-cell clustering. The algorithm is composed of two phases: in the initial phase, in order to ensure the high efficiency of sample random selection while maintaining the availability of k-means selection of sample representatives, a hybrid sample representative selection strategy is introduced; in the second phase, the K nearest representatives of any data object in the dataset are efficiently approximated by a rough-to-fine method, with a sparse affinity submatrix constructed between these objects and representatives. Then, the affinity submatrix is transformed into a bipartite graph, and the graph is effectively segmented by transfer cutting to achieve the clustering result. Finally, U-SENC integrates the previous multiple ultra-scalable spectral clustering (U-SPEC) to improve the robustness of the U-SPEC algorithm as well as keeping high effectiveness. The test results demonstrate that the clustering algorithm raised in this paper could better identify cell types from the aspect of clustering evaluation index normalized mutual information (NMI).

Integration of single sample and population analysis for understanding immune evasion mechanisms of lung cancer

Abstract: A deep understanding of the complex interaction mechanism between the various cellular components in tumor microenvironment (TME) of lung adenocarcinoma (LUAD) is a prerequisite for understanding its drug resistance, recurrence, and metastasis. In this study, we proposed two complementary computational frameworks for integrating multi-source and multi-omics data, namely ImmuCycReg framework (single sample level) and L0Reg framework (population or subtype level), to carry out difference analysis between the normal population and different LUAD subtypes. Then, we aimed to identify the possible immune escape pathways adopted by patients with different LUAD subtypes, resulting in immune deficiency which may occur at different stages of the immune cycle. More importantly, combining the research results of the single sample level and population level can improve the credibility of the regulatory network analysis results. In addition, we also established a prognostic scoring model based on the risk factors identified by Lasso-Cox method to predict survival of LUAD patients. The experimental results showed that our frameworks could reliably identify transcription factor (TF) regulating immune-related genes and could analyze the dominant immune escape pathways adopted by each LUAD subtype or even a single sample. Note that the proposed computational framework may be also applicable to the immune escape mechanism analysis of pan-cancer.

Annotating whole genome variants and constructing a multi-classifier based on samples of ADNI.

Abstract: INTRODUCTION Alzheimer's disease (AD) is the most common progressive neurodegenerative disorder in the elderly, which will eventually lead to dementia without an effective precaution and treatment. As a typical complex disease, the mechanism of AD's occurrence and development still lacks sufficient understanding. RESEARCH DESIGN AND METHODS In this study, we aim to directly analyze the relationship between DNA variants and phenotypes based on the whole genome sequencing data. Firstly, to enhance the biological meanings of our study, we annotate the deleterious variants and mapped them to nearest protein coding genes. Then, to eliminate the redundant features and reduce the burden of downstream analysis, a multi-objective evaluation strategy based on entropy theory is applied for ranking all candidate genes. Finally, we use multi-classifier XGBoost for classifying unbalanced data composed with 46 AD samples, 483 mild cognitive impairment (MCI) samples and 279 cognitive normal (CN) samples. RESULTS The experimental results on real whole genome sequencing data from Alzheimer's Disease Neuroimaging Initiative (ADNI) show that our method not only has satisfactory classification performance but also finds significance correlation between AD and RIN3, a known susceptibility gene of AD. In addition, pathway enrichment analysis was carried out using the top 20 feature genes, and three pathways were confirmed to be significantly related to the formation of AD. CONCLUSIONS From the experimental results, we demonstrated that the efficacy of our proposed method has practical significance.

Application of Graph Theory for Identifying Connectivity Patterns in Human Brain Networks: A Systematic Review.

Abstract: Background: Analysis of the human connectome using functional magnetic resonance imaging (fMRI) started in the mid-1990s and attracted increasing attention in attempts to discover the neural underpinnings of human cognition and neurological disorders. In general, brain connectivity patterns from fMRI data are classified as statistical dependencies (functional connectivity) or causal interactions (effective connectivity) among various neural units. Computational methods, especially graph theory-based methods, have recently played a significant role in understanding brain connectivity architecture. Objectives: Thanks to the emergence of graph theoretical analysis, the main purpose of the current paper is to systematically review how brain properties can emerge through the interactions of distinct neuronal units in various cognitive and neurological applications using fMRI. Moreover, this article provides an overview of the existing functional and effective connectivity methods used to construct the brain network, along with their advantages and pitfalls. Methods: In this systematic review, the databases Science Direct, Scopus, arXiv, Google Scholar, IEEE Xplore, PsycINFO, PubMed, and SpringerLink are employed for exploring the evolution of computational methods in human brain connectivity from 1990 to the present, focusing on graph theory. The Cochrane Collaboration's tool was used to assess the risk of bias in individual studies. Results: Our results show that graph theory and its implications in cognitive neuroscience have attracted the attention of researchers since 2009 (as the Human Connectome Project launched), because of their prominent capability in characterizing the behavior of complex brain systems. Although graph theoretical approach can be generally applied to either functional or effective connectivity patterns during rest or task performance, to date, most articles have focused on the resting-state functional connectivity. Conclusions: This review provides an insight into how to utilize graph theoretical measures to make neurobiological inferences regarding the mechanisms underlying human cognition and behavior as well as different brain disorders.