Article
Biochemistry & Molecular Biology
Robert Literman, Rachel Schwartz
Summary: Despite whole-genome sequencing data, evolutionary relationships remain controversial due to challenges in accurately modeling complex phylogenetic signals. Noncoding sequence sites provide more data and proportionally more concordant sites compared to coding sequences, which are also predominant drivers of tree incongruence.
MOLECULAR BIOLOGY AND EVOLUTION
(2021)
Article
Computer Science, Information Systems
Nephi A. Walton, Radha Nagarajan, Chen Wang, Murat Sincan, Robert R. Freimuth, David B. Everman, Derek C. Walton, Scott P. Mcgrath, Dominick J. Lemas, Panayiotis Benos, Alexander Alekseyenko, Qianqian Song, Ece Gamsiz Uzun, Casey Overby Taylor, Alper Uzun, Thomas Nate Person, Nadav Rappoport, Zhongming Zhao, Marc S. Williams
Summary: Substantial informatics research and development are needed to fully realize the clinical potential of AI in genomics. Developing larger datasets is crucial for emulating the success of AI in other domains. It is important to ensure that AI methods do not worsen existing socio-economic, racial, and ethnic disparities. Genomic data standards are critical for effectively scaling such technologies across institutions. The current focus should be on using these technologies in collaboration with clinicians, highlighting their value in clinical decision-making.
JOURNAL OF THE AMERICAN MEDICAL INFORMATICS ASSOCIATION
(2023)
Review
Biochemical Research Methods
Mattia Forcato, Oriana Romano, Silvio Bicciato
Summary: Recent advances in single-cell technologies are providing exciting opportunities for dissecting tissue heterogeneity and investigating cell identity, fate and function. However, combining different single-cell genomic signals is computationally challenging and requires integrative analysis.
BRIEFINGS IN BIOINFORMATICS
(2021)
Article
Computer Science, Information Systems
Navodini Wijethilake, Dulani Meedeniya, Charith Chitraranjan, Indika Perera, Mobarakol Islam, Hongliang Ren
Summary: Survival analysis plays a critical role in glioma patient management, with diverse approaches incorporating imaging and genetic information. Machine learning techniques and deep learning have emerged in recent years, replacing traditional statistical methods for survival analysis in glioma patients. Utilizing prognostic parameters acquired from diagnostic imaging techniques and genomic platforms is essential for survival or risk estimation of glioma patients.
Article
Computer Science, Information Systems
Ana Leon, Oscar Pastor
Summary: Understanding the human genome is a major research challenge that requires effective data management policies. Analyzing protein data can promote a shared understanding of the domain and facilitate the management of relevant genome data.
Article
Microbiology
Zhongyou Li, Katja Koeppen, Victoria I. Holden, Samuel L. Neff, Liviu Cengher, Elora G. Demers, Dallas L. Mould, Bruce A. Stanton, Thomas H. Hampton
Summary: Researchers developed an algorithm called GAUGE, which automatically annotates GEO microbial data sets, increasing the percentage of analyzable data sets from 4% to 33%. The annotations provide valuable insights and facilitate the identification of consistent patterns of differential gene expression. Additionally, they created a web interface called GAPE for reanalyzing P. aeruginosa and E. coli transcriptomic data.
Article
Microbiology
A. E. Roder, K. E. E. Johnson, M. Knoll, M. Khalfan, B. Wang, S. Schultz-Cherry, S. Banakis, A. Kreitman, C. Mederos, J. -H. Youn, R. Mercado, W. Wang, M. Chung, D. Ruchnewitz, M. I. Samanovic, M. J. Mulligan, M. Laessig, M. Luksza, S. Das, D. Gresham, E. Ghedin
Summary: High error rates in viral RNA-dependent RNA polymerases result in diverse populations of viruses within infected hosts. However, detecting minority variants in viral sequence data is challenging due to errors introduced during sample preparation and analysis. This study evaluated seven variant-calling tools using synthetic RNA controls and simulated data. The choice of variant caller and the use of replicate sequencing were found to have the greatest impact on single-nucleotide variant (SNV) discovery. The study provides guidance on identifying minority variants in SARS-CoV-2 clinical specimens and establishes heuristics for future studies on viral diversity and evolution.
Article
Biochemistry & Molecular Biology
Alexander M. Gout, Sasi Arunachalam, David B. Finkelstein, Jinghui Zhang
Summary: The review presents major data resources and actionable insights into the etiology of pediatric cancer, stemming from the identification of oncogenic gene fusions, mutational signature analysis, systems biology, and cancer predisposition studies, which have led to improved clinical diagnosis, discovery of new drug targets, and genetic predisposition screening. Integration of large-scale omics datasets through international collaboration is necessary to advance pediatric cancer research and clinical therapy.
BIOCHIMICA ET BIOPHYSICA ACTA-REVIEWS ON CANCER
(2021)
Article
Multidisciplinary Sciences
Elif Everest, Ege Ulgen, Ugur Uygunoglu, Melih Tutuncu, Sabahattin Saip, Osman Ugur Sezerman, Aksel Siva, Eda Tahir Turanli
Summary: By combining genomic and proteomic data of MS patients, shared pathways were identified, suggesting that integrating multiple datasets can reduce false positive and negative results in genome-wide SNP associations.
Article
Computer Science, Artificial Intelligence
Yongsheng Pan, Mingxia Liu, Yong Xia, Dinggang Shen
Summary: This study proposes a framework for joint neuroimage synthesis and disease diagnosis using incomplete multi-modality neuroimages. By designing a disease-image-specific network and a feature-consistency generative adversarial network, missing neuroimages can be synthesized while preserving disease-specific information. Experimental results show that the method achieves state-of-the-art performance in Alzheimer's disease identification and mild cognitive impairment conversion prediction.
IEEE TRANSACTIONS ON PATTERN ANALYSIS AND MACHINE INTELLIGENCE
(2022)
Article
Biochemistry & Molecular Biology
William Hemstrom, Melissa Jones
Summary: Genomic data analysis can be intimidating, especially for researchers without programming experience. snpR is a user-friendly R package that automates data subsetting and analysis based on categorical metadata, integrating methods from multiple packages to streamline repeated analyses.
MOLECULAR ECOLOGY RESOURCES
(2023)
Editorial Material
Microbiology
Taj Azarian
Summary: Population genomic analysis is a powerful tool to understand the evolutionary history and success factors of pathogens. This article demonstrates the utility of pathogen genomic data by elucidating the origin of methicillin-resistant Staphylococcus aureus ST239. The availability of representative genomic data and associated metadata is crucial for these analyses.
Article
Biology
Peiying Huang, Li Yan, Zhishang Li, Shuai Zhao, Yuchao Feng, Jing Zeng, Li Chen, Afang Huang, Yan Chen, Sisi Lei, Xiaoyan Huang, Yi Deng, Dan Xie, Hansu Guan, Weihang Peng, Liyuan Yu, Bojun Chen
Summary: This study aimed to identify shared gene signatures and related pathways between atherosclerosis and depression. Through differential and weighted gene co-expression network analyses, 165 atherosclerosis-related genes and 1478 depression-related genes were identified. Lipid and atherosclerosis pathway and tryptophan metabolism pathway were found to be potential pathways connecting atherosclerosis and depression. CASP1 and MMP9 were identified as critical crosstalk genes linking these two conditions.
COMPUTERS IN BIOLOGY AND MEDICINE
(2023)
Article
Computer Science, Interdisciplinary Applications
Wei Shao, Yingli Zuo, Yangyang Shi, Yawen Wu, Jiao Tang, Junyong Zhao, Liang Sun, Zixiao Lu, Jianpeng Sheng, Qi Zhu, Daoqiang Zhang
Summary: This study proposes an end-to-end deep learning framework, IMO-TILs, that integrates pathological images with multi-omics data to analyze tumor-infiltrating lymphocytes (TILs) and explore their survival-associated interactions with tumors. The framework utilizes a graph attention network to describe the spatial interactions between TILs and tumor regions in whole-slide pathological images, and applies a Concrete AutoEncoder (CAE) to select survival-associated genes from high-dimensional multi-omics data. Deep generalized canonical correlation analysis (DGCCA) and an attention layer are then used to fuse the image and multi-omics data for prognosis prediction of human cancers. The experimental results demonstrate that this method achieves higher prognosis results and identifies consistent imaging and multi-omics biomarkers strongly correlated with the prognosis of human cancers.
IEEE TRANSACTIONS ON MEDICAL IMAGING
(2023)
Article
Biology
Alexandra Danyi, Myrthe Jager, Jeroen de Ridder
Summary: This study introduces an improved method to address the sparsity in liquid biopsy data, achieving higher accuracy by data augmentation and integration. The results pave the way for the application of machine learning in detecting the cell of origin of cancer from liquid biopsy data.
Article
Computer Science, Interdisciplinary Applications
Li Guo, Jian Lyu, Zhe Zhang, Jinping Shi, Qianjin Feng, Yanqiu Feng, Mingyong Gao, Xinyuan Zhang
Summary: Diffusion kurtosis imaging (DKI) has shown its value in various applications, but accurate estimation of DKI tensors is often compromised by noise. This study proposes a joint denoising and estimating framework that integrates multiple sources of prior information to improve the estimation of DKI tensors. The results demonstrate that the proposed method outperforms other methods in simulations and in-vivo dMRI datasets with spatially stationary and nonstationary noise distributions. The study also confirms the effectiveness of integrating multiple sources of priors into the joint framework and the importance of local and nonlocal spatial smoothing constraints.
IEEE TRANSACTIONS ON MEDICAL IMAGING
(2022)
Article
Computer Science, Interdisciplinary Applications
Zhenyuan Ning, Shengzhou Zhong, Qianjin Feng, Wufan Chen, Yu Zhang
Summary: In this paper, a saliency-guided morphology-aware U-Net (SMU-Net) is proposed for lesion segmentation in breast ultrasound images. This method utilizes saliency maps to guide the network for learning foreground and background representations, and incorporates a middle stream for fusing features and enhancing morphological information. Experimental results demonstrate superior performance and robustness compared to state-of-the-art approaches.
IEEE TRANSACTIONS ON MEDICAL IMAGING
(2022)
Article
Computer Science, Artificial Intelligence
Zhichao Liang, Shuangyang Zhang, Jian Wu, Xipan Li, Zhijian Zhuang, Qianjin Feng, Wufan Chen, Li Qi
Summary: Preclinical imaging with photoacoustic tomography (PAT) has gained attention for its ability to provide molecular contrast with deep imaging depth. Automatic extraction and segmentation of animals in PAT images is crucial for improving image analysis efficiency and enabling advanced image post-processing. By proposing a volumetric auto-segmentation method based on the 3-D optimal graph search (3-D GS) algorithm, the study ensures surface continuity and enhances the accuracy and smoothness of segmented animal surfaces. Testing the method in vivo nude mice imaging experiments showed successful retention of continuous global surface structure and smooth local subcutaneous tumor boundaries in different development stages, leading to enhanced structural visibility and uniform image intensity in LF corrected PAT images.
MEDICAL IMAGE ANALYSIS
(2022)
Article
Computer Science, Artificial Intelligence
Shumao Pang, Chunlan Pang, Zhihai Su, Liyan Lin, Lei Zhao, Yangfan Chen, Yujia Zhou, Hai Lu, Qianjin Feng
Summary: The paper introduces a novel mixed-supervised segmentation network to reduce inter-class similarity in spine segmentation. By generating dynamic parameters for semantic feature map and utilizing a mixed-supervised loss, the network achieves state-of-the-art performance in automated spine segmentation.
MEDICAL IMAGE ANALYSIS
(2022)
Article
Engineering, Biomedical
Huixin Tan, Jiewei Lai, Yunbi Liu, Yuzhang Song, Jinliang Wang, Mingyang Chen, Yong Yan, Liming Zhong, Qianjin Feng, Wei Yang
Summary: In this study, a robust and lightweight quality assessment model for wearable ECG data is developed using neural architecture search algorithm. The model achieved excellent performance on large-scale datasets with high AUC and F1 scores. The proposed method is effective in real-time assessment of the quality of all leads of wearable ECG data on mobile devices.
BIOMEDICAL SIGNAL PROCESSING AND CONTROL
(2022)
Article
Orthopedics
Mifang Li, Hanhua Bai, Feiyuan Zhang, Yujia Zhou, Qiuyu Lin, Quan Zhou, Qianjin Feng, Lingyan Zhang
Summary: This study aimed to develop an MRI segmentation model of intercondylar fossa using deep learning to automatically measure notch volume and explore its correlation with ACL injury.
BMC MUSCULOSKELETAL DISORDERS
(2022)
Article
Computer Science, Interdisciplinary Applications
Zhaoqiang Yun, Qing Xu, Gengyuan Wang, Shuang Jin, Guoye Lin, Qianjin Feng, Jin Yuan
Summary: In this study, an EVA system based on deep learning was developed to quantitatively assess hemodynamics in conjunctival microvascular images. The system maintained vessel segmentation continuity and automatically measured blood velocity, providing an automatic and reliable solution.
COMPUTER METHODS AND PROGRAMS IN BIOMEDICINE
(2022)
Article
Neurosciences
Zehong Cao, Jiaona Xu, Bin Song, Lizhou Chen, Tianyang Sun, Yichu He, Ying Wei, Guozhong Niu, Yu Zhang, Qianjin Feng, Zhongxiang Ding, Feng Shi, Dinggang Shen
Summary: In this study, an automated ASPECTS scoring method utilizing neural networks was proposed. The method achieved remarkable performance in a large dataset and an independent testing dataset, and showed a high correlation between ASPECTS scores and patient prognosis.
HUMAN BRAIN MAPPING
(2022)
Article
Orthopedics
Zhihai Su, Zheng Liu, Min Wang, Shaolin Li, Liyan Lin, Zhen Yuan, Shumao Pang, Qianjin Feng, Tao Chen, Hai Lu
Summary: This study developed a new method for 3D reconstruction of Kambin's triangle using automated MRI segmentation. The method showed good performance in segmenting lumbar spinal structures and evaluating anatomical performance.
JOURNAL OF ORTHOPAEDIC RESEARCH
(2022)
Article
Computer Science, Artificial Intelligence
Xiaoxuan Zhang, Xiongfeng Zhu, Kai Tang, Yinghua Zhao, Zixiao Lu, Qianjin Feng
Summary: In this paper, a novel dense dual-task network (DDTNet) is proposed to achieve automatic detection and segmentation of tumor-infiltrating lymphocytes (TILs) in histopathological images. DDTNet utilizes a feature pyramid network for extracting multi-scale morphological characteristics of TILs, a detection module for locating TIL centers, and a segmentation module for delineating TIL boundaries. Experimental results show that DDTNet outperforms other methods in detection and segmentation metrics.
MEDICAL IMAGE ANALYSIS
(2022)
Article
Computer Science, Artificial Intelligence
Xiumei Chen, Tao Wang, Haoran Lai, Xiaoling Zhang, Qianjin Feng, Meiyan Huang
Summary: In this study, a novel method was proposed to analyze and detect the associations between multimodal imaging and genetic data for biomarker detection of neurodegenerative diseases. The method utilized structure constraints and nonlinear association analysis, achieving high accuracy of biomarker detection and validating previously identified disease-related biomarkers. This suggests that the method may provide insights into the pathological mechanism of neurodegenerative diseases and early prediction.
MEDICAL IMAGE ANALYSIS
(2022)
Article
Radiology, Nuclear Medicine & Medical Imaging
Guoye Lin, Hanhua Bai, Jie Zhao, Zhaoqiang Yun, Yangfan Chen, Shumao Pang, Qianjin Feng
Summary: The proposed method provides an accurate and robust solution for difficult vessel segmentation by introducing an error discrimination network to assist the segmentation network and training with different types of vessel samples and error masks, improving sensitivity to small vessels and the connectivity of segmentation results.
Article
Neurosciences
Hengbing Jiang, Lili Zou, Dequn Huang, Qianjin Feng
Summary: This article proposes and evaluates a novel method for continuous blood pressure estimation based on multi-scale feature extraction by a neural network with multi-task learning. The developed method achieves high accuracy and meets standard requirements without the need for calibration. It has the potential to enable continuous blood pressure monitoring by mobile health devices.
FRONTIERS IN NEUROSCIENCE
(2022)
Article
Computer Science, Artificial Intelligence
Zhenyuan Ning, Zehui Lin, Qing Xiao, Denghui Du, Qianjin Feng, Wufan Chen, Yu Zhang
Summary: In this article, a novel Cox-driven multi-constraint latent representation learning framework is proposed for prognosis analysis with multi-modal data. The framework efficiently fuses and selects complementary information from high-dimensional multi-modal data by learning a multi-modal latent space via a bi-mapping approach with ranking and regression constraints. The proposed method outperforms state-of-the-art Cox-based models according to extensive experiments on three datasets acquired from The Cancer Genome Atlas (TCGA).
IEEE TRANSACTIONS ON NEURAL NETWORKS AND LEARNING SYSTEMS
(2023)