Bioinspired Heterocoordination in Adaptable Cobalt Metal-Organic Framework for DNA Epigenetic Modification Detection.

Metal-organic frameworks (MOFs) show unique advantages in simulating the dynamics and fidelity of natural coordination. Inspired by zinc finger protein, a second linker was introduced to affect the homogeneous MOF system and thus facilitate the emergence of diverse functionalities. Under the systematic identification of 12 MOF species (i.e., metal ions, linkers) and 6 second linkers (trigger), a dissipative system consisting of Co-BDC-NO2 and o-phenylenediamine (oPD) was screened out, which can rapidly and in situ generate a high photothermal complex (η = 36.9%). Meanwhile, both the carboxylation of epigenetic modifications and metal ion (Fe3+, Ni2+, Cu2+, Zn2+, Co2+ and Mn2+) screening were utilized to improve the local coordination environment so that the adaptable Co-MOF growth on the DNA strand was realized. Thus, epigenetic modification information on DNA was converted to an amplified metal ion signal, and then oPD was further introduced to generate bimodal dissipative signals by which a simple, high-sensitivity detection strategy of 5-hydroxymethylcytosine (LOD = 0.02%) and 5-formylcytosine (LOD = 0.025‰) was developed. The strategy provides one low-cost method (< 0.01 $/sample) for quantifying global epigenetic modifications, which greatly promotes epigenetic modification-based early disease diagnosis. This work also proposes a general heterocoordination design concept for molecular recognition and signal transduction, opening a new MOF-based sensing paradigm.

Bioreaction-Compatible Bivariate Lanthanide MOF Sensor Enables Stimulus-Multiresponsive Platform for ctDNA On-Site Detection.

Detection of circulating tumor DNA (ctDNA) in liquid biopsy is of great importance for tumor diagnosis but difficult due to its low amount in bodily fluids. Herein, a novel ctDNA detection platform is established by quantifying DNA amplification by-product pyrophosphate (PPi) using a newly designed bivariable lanthanide metal-organic framework (Ln-MOF), namely, Ce/Eu-DPA MOF (CE-24, DPA = pyridine-2,6-dicarboxylic acid). CE-24 MOF exhibits ultrafast dual-response (fluorescence enhancement and enzyme-activity inhibition) to PPi stimuli by virtue of host-guest interaction. The platform is applied to detecting colon carcinoma-related ctDNA (KARS G12D mutation) combined with the isothermal nucleic acid exponential amplification reaction (EXPAR). ctDNA triggers the generation of a large amount of PPi, and the ctDNA quantification is achieved through the ratio fluorescence/colorimetric dual-mode assay of PPi. The combination of the EXPAR and the dual-mode PPi sensing allows the ctDNA assay method to be low-cost, convenient, bioreaction-compatible (freedom from the interference of bioreaction systems), sensitive (limit of detection down to 101 fM), and suitable for on-site detection. To the best of our knowledge, this work is the first application of Ln-MOF for ctDNA detection, and it provides a novel universal strategy for the rapid detection of nucleic acid biomarkers in point-of-care scenarios.

PT-KGNN: A framework for pre-training biomedical knowledge graphs with graph neural networks.

Biomedical knowledge graphs (KGs) serve as comprehensive data repositories that contain rich information about nodes and edges, providing modeling capabilities for complex relationships among biological entities. Many approaches either learn node features through traditional machine learning methods, or leverage graph neural networks (GNNs) to directly learn features of target nodes in the biomedical KGs and utilize them for downstream tasks. Motivated by the pre-training technique in natural language processing (NLP), we propose a framework named PT-KGNN (Pre-Training the biomedical KG with GNNs) to learn embeddings of nodes in a broader context by applying GNNs on the biomedical KG. We design several experiments to evaluate the effectivity of our proposed framework and the impact of the scale of KGs. The results of tasks consistently improve as the scale of the biomedical KG used for pre-training increases. Pre-training on large-scale biomedical KGs significantly enhances the drug-drug interaction (DDI) and drug-disease association (DDA) prediction performance on the independent dataset. The embeddings derived from a larger biomedical KG have demonstrated superior performance compared to those obtained from a smaller KG. By applying pre-training techniques on biomedical KGs, rich semantic and structural information can be learned, leading to enhanced performance on downstream tasks. it is evident that pre-training techniques hold tremendous potential and wide-ranging applications in bioinformatics.

Solvent-tuned perovskite heterostructures enable visual linoleic acid assay and edible oil species discrimination via wavelength shift.

Linoleic acid (LA) detection and edible oils discrimination are essential for food safety. Recently, CsPbBr3@SiO2 heterostructures have been widely applied in edible oil assays, while deep insights into solvent effects on their structure and performance are often overlooked. Based on the suitable polarity and viscosity of cyclohexane, we prepared CsPbBr3@SiO2 Janus nanoparticles (JNPs) with high stability in edible oil and fast halogen-exchange (FHE) efficiency with oleylammonium iodide (OLAI). LA is selectively oxidized by lipoxidase to yield hydroxylated derivative (oxLA) capable of reacting with OLAI, thereby bridging LA content to naked-eye fluorescence color changes through the anti-FHE reaction. The established method for LA in edible oils exhibited consistent results with GC-MS analysis (p > 0.05). Since the LA content difference between edible oils, we further utilized chemometrics to accurately distinguish (100%) the species of edible oils. Overall, such elaborated CsPbBr3@SiO2 JNPs enable a refreshing strategy for edible oil discrimination.

Hierarchical Self-Assembly Molecular Building Blocks as Intelligent Nanoplatforms for Ovarian Cancer Theranostics.

Hierarchical self-assembly from simple building blocks to complex polymers is a feasible approach to constructing multi-functional smart materials. However, the polymerization process of polymers often involves challenges such as the design of building blocks and the drive of external energy. Here, a hierarchical self-assembly with self-driven and energy conversion capabilities based on p-aminophenol and diethylenetriamine building blocks is reported. Through ß-galactosidase (ß-Gal) specific activation to the self-assembly, the intelligent assemblies (oligomer and superpolymer) with excellent photothermal and fluorescent properties are dynamically formed in situ, and thus the sensitive multi-mode detection of ß-Gal activity is realized. Based on the overexpression of ß-Gal in ovarian cancer cells, the self-assembly superpolymer is specifically generated in SKOV-3 cells to achieve fluorescence imaging. The photothermal therapeutic ability of the self-assembly oligomer (synthesized in vitro) is evaluated by a subcutaneous ovarian cancer model, showing satisfactory anti-tumor effects. This work expands the construction of intelligent assemblies through the self-driven cascade assembly of small molecules and provides new methods for the diagnosis and treatment of ovarian cancer.

Trinity Strategy: Enabling Perovskite as Hydrophilic and Efficient Fluorescent Nanozyme for Constructing Biomarker Reporting Platform.

Water instability and sensing homogeneity are the Achilles' heel of CsPbX3 NPs in biological fluids application. This work reports the preparation of Mn2+:CsPbCl3@SiO2 yolk-shell nanoparticles (YSNPs) in aqueous solutions created through the integration of ligand, surface, and crystal engineering strategies. The SN2 reaction between 4-chlorobutyric acid (CBA) and oleylamine (OAm) yields a zwitterionic ligand that facilitates the dispersion of YSNPs in water, while the robust SiO2 shell enhances their overall stability. Besides, Mn2+ doping in YSNPs not only introduces a second emission center but also enables potential postsynthetic designability, leading to the switching from YSNPs to MnO2@YSNPs with excellent oxidase (OXD)-like activity. Theoretical calculations reveal that electron transfer from CsPbCl3 to in situ MnO2 and the adsorption-desorption process of 3,3',5,5'-tetramethylbenzidine (TMB) synergistically amplify the OXD-like activity. In the presence of ascorbic acid (AA), Mn4+ in MnO2@YSNPs (fluorescent nanozyme) is reduced to Mn2+ and dissociated, thereby inhibiting the OXD-like activity and triggering fluorescence "turn-on/off", i.e., dual-mode recognition. Finally, a biomarker reporting platform based on MnO2@YSNPs fluorescent nanozyme is constructed with AA as the reporter molecule, and the accurate detection of human serum alkaline phosphatase (ALP) is realized, demonstrating the vast potential of perovskites in biosensing.

Macrophage-Inspired Degradation-Activation System Enables Ultrasensitive Multicolor Detection of Haloacetic Acids via Janus Perovskite Halide Exchange.

Haloacetic acids (HAAs), as representative disinfection byproducts, have the potential hazards of teratogenesis, carcinogenesis, and mutagenesis. Herein, inspired by the scavenging physiology of macrophages and taking advantage of the unique properties of perovskites, we design a biomimetic integrated three-step workflow, named the macrophage-inspired degradation-activation system (MIDAS), for the detection of HAAs in aqueous samples. First, HAAs are "devoured" by methyl t-butyl ether (MTBE) from a sample. Then, ultraviolet C is utilized to induce the photolysis of MTBE and the dehalogenation of HAAs. Third, the photoinduced product, tertiary butyl haloalkane, can activate the vacancy defect-facilitated halide exchange of perovskites, generating multicolor fluorescent signals. The MIDAS realizes the rapid (<5 min), ultrasensitive (limit of detection: 30 and 15 ppb), and accurate (recovery: 95.2-109.4%) quantification of dichloroacetic acid and dibromoacetic acid in real water samples. Furthermore, a chemometrics-supported MIDAS portable platform is established for the visual semi-quantification of HAAs and the discrimination of binary mixed HAAs on site. The MIDAS-based strategy provides a highly efficient approach to trigger the perovskite halide exchange and shows the Midas touch-like ability in the fluorescent assay of HAAs in aqueous samples. To our knowledge, this is the first universal multicolor fluorimetry and the first application of perovskites for HAA detection.

Visual Explanation for Open-domain Question Answering with BERT.

Open-domain question answering (OpenQA) is an essential but challenging task in natural language processing that aims to answer questions in natural language formats on the basis of large-scale unstructured passages. Recent research has taken the performance of benchmark datasets to new heights, especially when these datasets are combined with techniques for machine reading comprehension based on Transformer models. However, as identified through our ongoing collaboration with domain experts and our review of literature, three key challenges limit their further improvement: (i) complex data with multiple long texts, (ii) complex model architecture with multiple modules, and (iii) semantically complex decision process. In this paper, we present VEQA, a visual analytics system that helps experts understand the decision reasons of OpenQA and provides insights into model improvement. The system summarizes the data flow within and between modules in the OpenQA model as the decision process takes place at the summary, instance and candidate levels. Specifically, it guides users through a summary visualization of dataset and module response to explore individual instances with a ranking visualization that incorporates context. Furthermore, VEQA supports fine-grained exploration of the decision flow within a single module through a comparative tree visualization. We demonstrate the effectiveness of VEQA in promoting interpretability and providing insights into model enhancement through a case study and expert evaluation.

Measuring spatial accessibility of residents' medical treatment under hierarchical diagnosis and treatment system: Multi-scenario simulation in China.

In order to improve the operational efficiency of medical institutions and build a more complete and efficient medical system, the Chinese government is vigorously promoting the reform of hierarchical diagnosis and treatment. We constructed a multi-factor composite selection weight to characterize the residents' medical treatment behavior in the context of hierarchical diagnosis and treatment. By combining the weight with the two-step floating catchment area method, we analyzed the spatial variation characteristics of residents' accessibility to medical care under different scenarios. Results show that the referral rate between medical institutions increases gradually along with the occurrence of public health events. When there is a major public health event, the proportion of the population transferred from the primary medical institutions to the county hospitals and the county hospitals to the municipal hospitals exceeded 65%. In three scenarios, the spatial pattern of accessibility shows obvious consistency and local differences. Among the three-tier medical institutions in China, the service capacity of county hospitals is poor, and the contribution rate of accessibility is less than 20%. The results clearly show the spatial differences in the accessibility of Chinese residents in different scenarios and the impact of public health events on accessibility. This research can provide a reference for the layout optimization of medical resources in the future.

Optimization referral rate design for hierarchical diagnosis and treatment system based on accessibility-utilization efficiency bi-objective collaboration: A case study of China.

The hierarchical diagnosis and treatment reform of China can guide residents to seek medical treatment in an orderly manner and improve access to medical treatment. Most existing studies on hierarchical diagnosis and treatment used accessibility as the evaluation index to determine the referral rate between hospitals. However, the blind pursuit of accessibility will cause the problem of uneven utilization efficiency of hospitals at different levels. In response to this, we constructed a bi-objective optimization model based on the perspective of residents and medical institutions. This model can give the optimal referral rate for each province considering the accessibility of residents and the utilization efficiency of hospitals, to improve the utilization efficiency and equality of access for hospitals. The results showed that the applicability of bi-objective optimization model is good, and the optimal referral rate based on the model can ensure the maximum benefit of the two optimization goals. In the optimal referral rate model, residents' medical accessibility is relatively balanced overall. In terms of obtaining high-grade medical resources, the accessibility is better in the eastern and central regions, but poorer in the western China. According to the current allocation of medical resources in China, the medical tasks undertaken by high-grade hospitals account for 60%-78%, which are still the main force of medical services. In this way, there is a big gap in realizing the "serious diseases do not leave the county" goal of hierarchical diagnosis and treatment reform.

A benchmark for automatic medical consultation system: frameworks, tasks and datasets.

MOTIVATION: In recent years, interest has arisen in using machine learning to improve the efficiency of automatic medical consultation and enhance patient experience. In this article, we propose two frameworks to support automatic medical consultation, namely doctor-patient dialogue understanding and task-oriented interaction. We create a new large medical dialogue dataset with multi-level fine-grained annotations and establish five independent tasks, including named entity recognition, dialogue act classification, symptom label inference, medical report generation and diagnosis-oriented dialogue policy. RESULTS: We report a set of benchmark results for each task, which shows the usability of the dataset and sets a baseline for future studies. AVAILABILITY AND IMPLEMENTATION: Both code and data are available from https://github.com/lemuria-wchen/imcs21. SUPPLEMENTARY INFORMATION: Supplementary data are available at Bioinformatics online.

DxFormer: a decoupled automatic diagnostic system based on decoder-encoder transformer with dense symptom representations.

MOTIVATION: Symptom-based automatic diagnostic system queries the patient's potential symptoms through continuous interaction with the patient and makes predictions about possible diseases. A few studies use reinforcement learning (RL) to learn the optimal policy from the joint action space of symptoms and diseases. However, existing RL (or Non-RL) methods focus on disease diagnosis while ignoring the importance of symptom inquiry. Although these systems have achieved considerable diagnostic accuracy, they are still far below its performance upper bound due to few turns of interaction with patients and insufficient performance of symptom inquiry. To address this problem, we propose a new automatic diagnostic framework called DxFormer, which decouples symptom inquiry and disease diagnosis, so that these two modules can be independently optimized. The transition from symptom inquiry to disease diagnosis is parametrically determined by the stopping criteria. In DxFormer, we treat each symptom as a token, and formalize the symptom inquiry and disease diagnosis to a language generation model and a sequence classification model, respectively. We use the inverted version of Transformer, i.e. the decoder-encoder structure, to learn the representation of symptoms by jointly optimizing the reinforce reward and cross-entropy loss. RESULTS: We conduct experiments on three real-world medical dialogue datasets, and the experimental results verify the feasibility of increasing diagnostic accuracy by improving symptom recall. Our model overcomes the shortcomings of previous RL-based methods. By decoupling symptom query from the process of diagnosis, DxFormer greatly improves the symptom recall and achieves the state-of-the-art diagnostic accuracy. AVAILABILITY AND IMPLEMENTATION: Both code and data are available at https://github.com/lemuria-wchen/DxFormer. SUPPLEMENTARY INFORMATION: Supplementary data are available at Bioinformatics online.

In Situ Formation of o-Phenylenediamine Cascade Polymers Mediated by Metal-Organic Framework Nanozymes for Fluorescent and Photothermal Dual-Mode Assay of Acetylcholinesterase Activity.

A fluorescent and photothermal dual-mode assay method was established for the detection of acetylcholinesterase (AChE) activity based on in situ formation of o-phenylenediamine (oPD) cascade polymers. First, copper metal-organic frameworks of benzenetricarboxylic acid (Cu-BTC) were screened out as nanozymes with excellent oxidase-like activity and confinement catalysis effect. Then, an ingenious oPD cascade polymerization strategy was proposed. That is, oPD was oxidized by Cu-BTC to oPD oligomers with strong yellow fluorescence, and oPD oligomers were further catalyzed to generate J-aggregation, which promotes the formation of oPD polymer nanoparticles with a high photothermal effect. By utilizing thiocholine (enzymolysis product of acetylthiocholine) to inhibit the Cu-BTC catalytic effect, AChE activity was detected through the fluorescence-photothermal dual-signal change of oPD oligomers and polymer nanoparticles. Both assay modes have low detection limitation (0.03 U L-1 for fluorescence and 0.05 U L-1 for photothermal) and can accurately detect the AChE activity of human serum (recovery 85.0-111.3%). The detection results of real serum samples by fluorescent and photothermal dual modes are consistent with each other (relative error ≤ 5.2%). It is worth emphasizing that this is the first time to report the high photothermal effect of oPD polymers and the fluorescence-photothermal dual-mode assay of enzyme activity.

A review and performance evaluation of clustering frameworks for single-cell Hi-C data.

The three-dimensional genome structure plays a key role in cellular function and gene regulation. Single-cell Hi-C (high-resolution chromosome conformation capture) technology can capture genome structure information at the cell level, which provides the opportunity to study how genome structure varies among different cell types. Recently, a few methods are well designed for single-cell Hi-C clustering. In this manuscript, we perform an in-depth benchmark study of available single-cell Hi-C data clustering methods to implement an evaluation system for multiple clustering frameworks based on both human and mouse datasets. We compare eight methods in terms of visualization and clustering performance. Performance is evaluated using four benchmark metrics including adjusted rand index, normalized mutual information, homogeneity and Fowlkes-Mallows index. Furthermore, we also evaluate the eight methods for the task of separating cells at different stages of the cell cycle based on single-cell Hi-C data.

Hierarchical reinforcement learning for automatic disease diagnosis.

MOTIVATION: Disease diagnosis-oriented dialog system models the interactive consultation procedure as the Markov decision process, and reinforcement learning algorithms are used to solve the problem. Existing approaches usually employ a flat policy structure that treat all symptoms and diseases equally for action making. This strategy works well in a simple scenario when the action space is small; however, its efficiency will be challenged in the real environment. Inspired by the offline consultation process, we propose to integrate a hierarchical policy structure of two levels into the dialog system for policy learning. The high-level policy consists of a master model that is responsible for triggering a low-level model, the low-level policy consists of several symptom checkers and a disease classifier. The proposed policy structure is capable to deal with diagnosis problem including large number of diseases and symptoms. RESULTS: Experimental results on three real-world datasets and a synthetic dataset demonstrate that our hierarchical framework achieves higher accuracy and symptom recall in disease diagnosis compared with existing systems. We construct a benchmark including datasets and implementation of existing algorithms to encourage follow-up researches. AVAILABILITY AND IMPLEMENTATION: The code and data are available from https://github.com/FudanDISC/DISCOpen-MedBox-DialoDiagnosis. SUPPLEMENTARY INFORMATION: Supplementary data are available at Bioinformatics online.

Portable visual assay of Bacillus anthracis biomarker based on ligand-functionalized dual-emission lanthanide metal-organic frameworks and smartphone-integrated mini-device.

A single-functionalized ligand single-Ln3+ based dual-emission Ln-MOF fluorescent sensor was established for portable and visual dipicolinic acid (DPA, Bacillus anthracis biomarker) detection. First, a theory calculation-based prediction model was developed for designing single-functionalized ligand single-Ln3+ dual-emission Ln-MOFs. The model consisted of three calculation modules: intramolecular hydrogen bonds, excited state energy levels, and coordination stability with Ln3+ of ligands. Tb3+ and Eu3+ were selected as metal luminescence centers, PTA-X (PTA: p-phthalic acid, X = NH2, CH3, H, OH) with different functional groups as one-step functionalization ligands, and the luminescent feature of four Tb-MOFs and four Eu-MOFs was predicted with the model. Coupled with prediction results and experimental verification results, Tb-PTA-OH was rapidly determined to be the sole dual-emission Ln-MOF. Then, Tb-PTA-OH was applied to DPA detection by ratiometric fluorescence, and an ultra-low limit of detection (13.4 nM) was obtained, which is much lower than the lowest anthrax infectious dose (60 µM). A portable visual assay method based on paper-microchip and smartphone integrated mini-device was further established (limit of qualification 0.48 µM). A new sensing mechanism and a "triple gates" selectivity mechanism to DPA were proposed. This work reveals guidelines for material design and mini-device customization in detecting hazardous substances.

Inferring Association Between Alcohol Addiction and Defendant's Emotion Based on Sound at Court.

Alcohol addiction can lead to health and social problems. It can also affect people's emotions. Emotion plays a key role in human communications. It is important to recognize the people's emotions at the court and infer the association between the people's emotions and the alcohol addiction. However, it is challenging to recognize people's emotions efficiently in the courtroom. Furthermore, to the best of our knowledge, no existing work is about the association between alcohol addiction and people's emotions at court. In this paper, we propose a deep learning framework for predicting people's emotions based on sound perception, named ResCNN-SER. The proposed model combines several neural network-based components to extract the features of the speech signals and predict the emotions. The evaluation shows that the proposed model performs better than existing methods. By applying ResCNN-SER for emotion recognition based on people's voices at court, we infer the association between alcohol addiction and the defendant's emotion at court. Based on the sound source data from 54 trial records, we found that the defendants with alcohol addiction tend to get angry or fearful more easily at court comparing with defendants without alcohol addiction.

An end-to-end heterogeneous graph representation learning-based framework for drug-target interaction prediction.

Accurately identifying potential drug-target interactions (DTIs) is a key step in drug discovery. Although many related experimental studies have been carried out for identifying DTIs in the past few decades, the biological experiment-based DTI identification is still timeconsuming and expensive. Therefore, it is of great significance to develop effective computational methods for identifying DTIs. In this paper, we develop a novel 'end-to-end' learning-based framework based on heterogeneous 'graph' convolutional networks for 'DTI' prediction called end-to-end graph (EEG)-DTI. Given a heterogeneous network containing multiple types of biological entities (i.e. drug, protein, disease, side-effect), EEG-DTI learns the low-dimensional feature representation of drugs and targets using a graph convolutional networks-based model and predicts DTIs based on the learned features. During the training process, EEG-DTI learns the feature representation of nodes in an end-to-end mode. The evaluation test shows that EEG-DTI performs better than existing state-of-art methods. The data and source code are available at: https://github.com/MedicineBiology-AI/EEG-DTI.

Integrating multi-network topology for gene function prediction using deep neural networks.

MOTIVATION: The emergence of abundant biological networks, which benefit from the development of advanced high-throughput techniques, contributes to describing and modeling complex internal interactions among biological entities such as genes and proteins. Multiple networks provide rich information for inferring the function of genes or proteins. To extract functional patterns of genes based on multiple heterogeneous networks, network embedding-based methods, aiming to capture non-linear and low-dimensional feature representation based on network biology, have recently achieved remarkable performance in gene function prediction. However, existing methods do not consider the shared information among different networks during the feature learning process. RESULTS: Taking the correlation among the networks into account, we design a novel semi-supervised autoencoder method to integrate multiple networks and generate a low-dimensional feature representation. Then we utilize a convolutional neural network based on the integrated feature embedding to annotate unlabeled gene functions. We test our method on both yeast and human datasets and compare with three state-of-the-art methods. The results demonstrate the superior performance of our method. We not only provide a comprehensive analysis of the performance of the newly proposed algorithm but also provide a tool for extracting features of genes based on multiple networks, which can be used in the downstream machine learning task. AVAILABILITY: DeepMNE-CNN is freely available at https://github.com/xuehansheng/DeepMNE-CNN. CONTACT: jiajiepeng@nwpu.edu.cn; shang@nwpu.edu.cn; jianye.hao@tju.edu.cn.