DP-SSLoRA: A privacy-preserving medical classification model combining differential privacy with self-supervised low-rank adaptation.

BACKGROUND AND OBJECTIVE: Concerns about patient privacy issues have limited the application of medical deep learning models in certain real-world scenarios. Differential privacy (DP) can alleviate this problem by injecting random noise into the model. However, naively applying DP to medical models will not achieve a satisfactory balance between privacy and utility due to the high dimensionality of medical models and the limited labeled samples. METHODS: This work proposed the DP-SSLoRA model, a privacy-preserving classification model for medical images combining differential privacy with self-supervised low-rank adaptation. In this work, a self-supervised pre-training method is used to obtain enhanced representations from unlabeled publicly available medical data. Then, a low-rank decomposition method is employed to mitigate the impact of differentially private noise and combined with pre-trained features to conduct the classification task on private datasets. RESULTS: In the classification experiments using three real chest-X ray datasets, DP-SSLoRA achieves good performance with strong privacy guarantees. Under the premise of ɛ=2, with the AUC of 0.942 in RSNA, the AUC of 0.9658 in Covid-QU-mini, and the AUC of 0.9886 in Chest X-ray 15k. CONCLUSION: Extensive experiments on real chest X-ray datasets show that DP-SSLoRA can achieve satisfactory performance with stronger privacy guarantees. This study provides guidance for studying privacy-preserving in the medical field. Source code is publicly available online. https://github.com/oneheartforone/DP-SSLoRA.

MPASL: multi-perspective learning knowledge graph attention network for synthetic lethality prediction in human cancer.

Synthetic lethality (SL) is widely used to discover the anti-cancer drug targets. However, the identification of SL interactions through wet experiments is costly and inefficient. Hence, the development of efficient and high-accuracy computational methods for SL interactions prediction is of great significance. In this study, we propose MPASL, a multi-perspective learning knowledge graph attention network to enhance synthetic lethality prediction. MPASL utilizes knowledge graph hierarchy propagation to explore multi-source neighbor nodes related to genes. The knowledge graph ripple propagation expands gene representations through existing gene SL preference sets. MPASL can learn the gene representations from both gene-entity perspective and entity-entity perspective. Specifically, based on the aggregation method, we learn to obtain gene-oriented entity embeddings. Then, the gene representations are refined by comparing the various layer-wise neighborhood features of entities using the discrepancy contrastive technique. Finally, the learned gene representation is applied in SL prediction. Experimental results demonstrated that MPASL outperforms several state-of-the-art methods. Additionally, case studies have validated the effectiveness of MPASL in identifying SL interactions between genes.

Drug-drug interactions prediction based on deep learning and knowledge graph: A review.

Drug-drug interactions (DDIs) can produce unpredictable pharmacological effects and lead to adverse events that have the potential to cause irreversible damage to the organism. Traditional methods to detect DDIs through biological or pharmacological analysis are time-consuming and expensive, therefore, there is an urgent need to develop computational methods to effectively predict drug-drug interactions. Currently, deep learning and knowledge graph techniques which can effectively extract features of entities have been widely utilized to develop DDI prediction methods. In this research, we aim to systematically review DDI prediction researches applying deep learning and graph knowledge. The available biomedical data and public databases related to drugs are firstly summarized in this review. Then, we discuss the existing drug-drug interactions prediction methods which have utilized deep learning and knowledge graph techniques and group them into three main classes: deep learning-based methods, knowledge graph-based methods, and methods that combine deep learning with knowledge graph. We comprehensively analyze the commonly used drug related data and various DDI prediction methods, and compare these prediction methods on benchmark datasets. Finally, we briefly discuss the challenges related to drug-drug interactions prediction, including asymmetric DDIs prediction and high-order DDI prediction.

Prediction of drug-disease associations based on reinforcement symmetric metric learning and graph convolution network.

Accurately identifying novel indications for drugs is crucial in drug research and discovery. Traditional drug discovery is costly and time-consuming. Computational drug repositioning can provide an effective strategy for discovering potential drug-disease associations. However, the known experimentally verified drug-disease associations is relatively sparse, which may affect the prediction performance of the computational drug repositioning methods. Moreover, while the existing drug-disease prediction method based on metric learning algorithm has achieved better performance, it simply learns features of drugs and diseases only from the drug-centered perspective, and cannot comprehensively model the latent features of drugs and diseases. In this study, we propose a novel drug repositioning method named RSML-GCN, which applies graph convolutional network and reinforcement symmetric metric learning to predict potential drug-disease associations. RSML-GCN first constructs a drug-disease heterogeneous network by integrating the association and feature information of drugs and diseases. Then, the graph convolutional network (GCN) is applied to complement the drug-disease association information. Finally, reinforcement symmetric metric learning with adaptive margin is designed to learn the latent vector representation of drugs and diseases. Based on the learned latent vector representation, the novel drug-disease associations can be identified by the metric function. Comprehensive experiments on benchmark datasets demonstrated the superior prediction performance of RSML-GCN for drug repositioning.

KGANSynergy: knowledge graph attention network for drug synergy prediction.

Combination therapy is widely used to treat complex diseases, particularly in patients who respond poorly to monotherapy. For example, compared with the use of a single drug, drug combinations can reduce drug resistance and improve the efficacy of cancer treatment. Thus, it is vital for researchers and society to help develop effective combination therapies through clinical trials. However, high-throughput synergistic drug combination screening remains challenging and expensive in the large combinational space, where an array of compounds are used. To solve this problem, various computational approaches have been proposed to effectively identify drug combinations by utilizing drug-related biomedical information. In this study, considering the implications of various types of neighbor information of drug entities, we propose a novel end-to-end Knowledge Graph Attention Network to predict drug synergy (KGANSynergy), which utilizes neighbor information of known drugs/cell lines effectively. KGANSynergy uses knowledge graph (KG) hierarchical propagation to find multi-source neighbor nodes for drugs and cell lines. The knowledge graph attention network is designed to distinguish the importance of neighbors in a KG through a multi-attention mechanism and then aggregate the entity's neighbor node information to enrich the entity. Finally, the learned drug and cell line embeddings can be utilized to predict the synergy of drug combinations. Experiments demonstrated that our method outperformed several other competing methods, indicating that our method is effective in identifying drug combinations.

SLHSD: hybrid scaffolding method based on short and long reads.

In genome assembly, scaffolding can obtain more complete and continuous scaffolds. Current scaffolding methods usually adopt one type of read to construct a scaffold graph and then orient and order contigs. However, scaffolding with the strengths of two or more types of reads seems to be a better solution to some tricky problems. Combining the advantages of different types of data is significant for scaffolding. Here, a hybrid scaffolding method (SLHSD) is present that simultaneously leverages the precision of short reads and the length advantage of long reads. Building an optimal scaffold graph is an important foundation for getting scaffolds. SLHSD uses a new algorithm that combines long and short read alignment information to determine whether to add an edge and how to calculate the edge weight in a scaffold graph. In addition, SLHSD develops a strategy to ensure that edges with high confidence can be added to the graph with priority. Then, a linear programming model is used to detect and remove remaining false edges in the graph. We compared SLHSD with other scaffolding methods on five datasets. Experimental results show that SLHSD outperforms other methods. The open-source code of SLHSD is available at https://github.com/luojunwei/SLHSD.

Double-Sided Sapphire Optrodes with Conductive Shielding Layers to Reduce Optogenetic Stimulation Artifacts.

Optrodes, which are single shaft neural probes integrated with microelectrodes and optical light sources, offer a remarkable opportunity to simultaneously record and modulate neural activities using light within an animal's brain; however, a common problem with optrodes is that stimulation artifacts can be observed in the neural recordings of microelectrodes when the light source on the optrode is activated. These stimulation artifacts are undesirable contaminants, and they cause interpretation complexity when analyzing the recorded neural activities. In this paper, we tried to mitigate the effects of the stimulation artifacts by developing a low-noise, double-sided optrode integrated with multiple Electromagnetic Shielding (EMS) layers. The LED and microelectrodes were constructed separately on the top epitaxial and bottom substrate layers, and EMS layers were used to separate the microelectrodes and LED to reduce signal cross-talks. Compared with conventional single-sided designs, in which the LED and microelectrodes are constructed on the same side, our results indicate that double-sided optrodes can significantly reduce the presence of stimulation artifacts. In addition, the presence of stimulation artifacts can further be reduced by decreasing the voltage difference and increasing the rise/fall time of the driving LED pulsed voltage. With all these strategies, the presence of stimulation artifacts was significantly reduced by ~76%. As well as stimulation suppression, the sapphire substrate also provided strong mechanical stiffness and support to the optrodes, as well as improved electronic stability, thus making the double-sided sapphire optrodes highly suitable for optogenetic neuroscience research on animal models.

Deep learning approach for cancer subtype classification using high-dimensional gene expression data.

MOTIVATION: Studies have shown that classifying cancer subtypes can provide valuable information for a range of cancer research, from aetiology and tumour biology to prognosis and personalized treatment. Current methods usually adopt gene expression data to perform cancer subtype classification. However, cancer samples are scarce, and the high-dimensional features of their gene expression data are too sparse to allow most methods to achieve desirable classification results. RESULTS: In this paper, we propose a deep learning approach by combining a convolutional neural network (CNN) and bidirectional gated recurrent unit (BiGRU): our approach, DCGN, aims to achieve nonlinear dimensionality reduction and learn features to eliminate irrelevant factors in gene expression data. Specifically, DCGN first uses the synthetic minority oversampling technique algorithm to equalize data. The CNN can handle high-dimensional data without stress and extract important local features, and the BiGRU can analyse deep features and retain their important information; the DCGN captures key features by combining both neural networks to overcome the challenges of small sample sizes and sparse, high-dimensional features. In the experiments, we compared the DCGN to seven other cancer subtype classification methods using breast and bladder cancer gene expression datasets. The experimental results show that the DCGN performs better than the other seven methods and can provide more satisfactory classification results.

120 GHz Frequency-Doubler Module Based on GaN Schottky Barrier Diode.

Traditional GaAs-based frequency multipliers still exhibit great challenges to meet the demand for solid-state high-power THz sources due to low breakdown voltage and heat dissipation of the Schottky barrier diode (SBD). In this study, a GaN SBD chain was fabricated with n-/n+-GaN structure. As a consequence, the breakdown voltage of 54.9 V at 1 µA and cut-off frequency of 587.5 GHz at zero bias were obtained. A 120 GHz frequency-doubler module based on the GaN SBD chain was designed and fabricated. When driven with 500 mW input power in a continuous wave, the output power of the frequency-doubler module was 15.1 mW at 120 GHz. Moreover, the experiments show that the frequency-doubler module can endure an input power of 2 W. In addition, it is worth noting that the SBD chain works well at an anode temperature of 337.2 °C.

DACPGTN: Drug ATC Code Prediction Method Based on Graph Transformer Network for Drug Discovery.

The Anatomical Therapeutic Chemical (ATC) classification system is a drug classification scheme proposed by the World Health Organization, which is widely used for drug screening, repositioning, and similarity research. The ATC system assigns different ATC codes to drugs based on their anatomy, pharmacological, therapeutics and chemical properties. Predicting the ATC code of a given drug helps to understand the indication and potential toxicity of the drug, thus promoting its use in the therapeutic phase and accelerating its development. In this article, we propose an end-to-end model DACPGTN to predict the ATC code for the given drug. DACPGTN constructs composite features of drugs, diseases and targets by applying diverse biomedical information. Inspired by the application of Graph Transformer Network, we learn potential novel interactions among drugs diseases and targets from the known interactions to construct drug-target-disease heterogeneous networks containing comprehensive interaction information. Based on the constructed composite features and learned heterogeneous networks, we employ graph convolution network to generate the embedding of drug nodes, which are further used for the multi-label learning tasks in drug discovery. Experiments on the benchmark datasets demonstrate that the proposed DACPGTN model can achieve better prediction performance than the existing methods. The source codes of our method are available at https://github.com/Szhgege/DACPGTN.

A novel liver cancer diagnosis method based on patient similarity network and DenseGCN.

Liver cancer is the main malignancy in terms of mortality rate, accurate diagnosis can help the treatment outcome of liver cancer. Patient similarity network is an important information which helps in cancer diagnosis. However, recent works rarely take patient similarity into consideration. To address this issue, we constructed patient similarity network using three liver cancer omics data, and proposed a novel liver cancer diagnosis method consisted of similarity network fusion, denoising autoencoder and dense graph convolutional neural network to capitalize on patient similarity network and multi omics data. We compared our proposed method with other state-of-the-art methods and machine learning methods on TCGA-LIHC dataset to evaluate its performance. The results confirmed that our proposed method surpasses these comparison methods in terms of all the metrics. Especially, our proposed method has attained an accuracy up to 0.9857.

MultiGATAE: A Novel Cancer Subtype Identification Method Based on Multi-Omics and Attention Mechanism.

Cancer is one of the leading causes of death worldwide, which brings an urgent need for its effective treatment. However, cancer is highly heterogeneous, meaning that one cancer can be divided into several subtypes with distinct pathogenesis and outcomes. This is considered as the main problem which limits the precision treatment of cancer. Thus, cancer subtypes identification is of great importance for cancer diagnosis and treatment. In this work, we propose a deep learning method which is based on multi-omics and attention mechanism to effectively identify cancer subtypes. We first used similarity network fusion to integrate multi-omics data to construct a similarity graph. Then, the similarity graph and the feature matrix of the patient are input into a graph autoencoder composed of a graph attention network and omics-level attention mechanism to learn embedding representation. The K-means clustering method is applied to the embedding representation to identify cancer subtypes. The experiment on eight TCGA datasets confirmed that our proposed method performs better for cancer subtypes identification when compared with the other state-of-the-art methods. The source codes of our method are available at https://github.com/kataomoi7/multiGATAE.

BreakNet: detecting deletions using long reads and a deep learning approach.

BACKGROUND: Structural variations (SVs) occupy a prominent position in human genetic diversity, and deletions form an important type of SV that has been suggested to be associated with genetic diseases. Although various deletion calling methods based on long reads have been proposed, a new approach is still needed to mine features in long-read alignment information. Recently, deep learning has attracted much attention in genome analysis, and it is a promising technique for calling SVs. RESULTS: In this paper, we propose BreakNet, a deep learning method that detects deletions by using long reads. BreakNet first extracts feature matrices from long-read alignments. Second, it uses a time-distributed convolutional neural network (CNN) to integrate and map the feature matrices to feature vectors. Third, BreakNet employs a bidirectional long short-term memory (BLSTM) model to analyse the produced set of continuous feature vectors in both the forward and backward directions. Finally, a classification module determines whether a region refers to a deletion. On real long-read sequencing datasets, we demonstrate that BreakNet outperforms Sniffles, SVIM and cuteSV in terms of their F1 scores. The source code for the proposed method is available from GitHub at https://github.com/luojunwei/BreakNet . CONCLUSIONS: Our work shows that deep learning can be combined with long reads to call deletions more effectively than existing methods.

MI_DenseNetCAM: A Novel Pan-Cancer Classification and Prediction Method Based on Mutual Information and Deep Learning Model.

The Pan-Cancer Atlas consists of original sequencing data from various sources, provides the opportunity to perform systematic studies on the commonalities and differences between diverse cancers. The analysis for the pan-cancer dataset could help researchers to identify the key factors that could trigger cancer. In this paper, we present a novel pan-cancer classification method, referred to MI_DenseNetCAM, to identify a set of genes that can differentiate all tumor types accurately. First, the Mutual Information (MI) was utilized to eliminate noise and redundancy from the pan-cancer datasets. Then, the gene data was further converted to 2D images. Next, the DenseNet model was adopted as a classifier and the Guided Grad-CAM algorithm was applied to identify the key genes. Extensive experimental results on the public RNA-seq data sets with 33 different tumor types show that our method outperforms the other state-of-the-art classification methods. Moreover, gene analysis further demonstrated that the genes selected by our method were related to the corresponding tumor types.

Predicting Drug-Disease Association Based on Ensemble Strategy.

Drug repositioning is used to find new uses for existing drugs, effectively shortening the drug research and development cycle and reducing costs and risks. A new model of drug repositioning based on ensemble learning is proposed. This work develops a novel computational drug repositioning approach called CMAF to discover potential drug-disease associations. First, for new drugs and diseases or unknown drug-disease pairs, based on their known neighbor information, an association probability can be obtained by implementing the weighted K nearest known neighbors (WKNKN) method and improving the drug-disease association information. Then, a new drug similarity network and new disease similarity network can be constructed. Three prediction models are applied and ensembled to enable the final association of drug-disease pairs based on improved drug-disease association information and the constructed similarity network. The experimental results demonstrate that the developed approach outperforms recent state-of-the-art prediction models. Case studies further confirm the predictive ability of the proposed method. Our proposed method can effectively improve the prediction results.

SIns: A Novel Insertion Detection Approach Based on Soft-Clipped Reads.

As a common type of structural variation, an insertion refers to the addition of a DNA sequence into an individual genome and is usually associated with some inherited diseases. In recent years, many methods have been proposed for detecting insertions. However, the accurate calling of insertions is also a challenging task. In this study, we propose a novel insertion detection approach based on soft-clipped reads, which is called SIns. First, based on the alignments between paired reads and the reference genome, SIns extracts breakpoints from soft-clipped reads and determines insertion locations. The insert size information about paired reads is then further clustered to determine the genotype, and SIns subsequently adopts Minia to assemble the insertion sequences. Experimental results show that SIns can achieve better performance than other methods in terms of the F-score value for simulated and true datasets.

A Novel Biomarker Identification Approach for Gastric Cancer Using Gene Expression and DNA Methylation Dataset.

As one type of complex disease, gastric cancer has high mortality rate, and there are few effective treatments for patients in advanced stage. With the development of biological technology, a large amount of multiple-omics data of gastric cancer are generated, which enables computational method to discover potential biomarkers of gastric cancer. That will be very important to detect gastric cancer at earlier stages and thus assist in providing timely treatment. However, most of biological data have the characteristics of high dimension and low sample size. It is hard to process directly without feature selection. Besides, only using some omic data, such as gene expression data, provides limited evidence to investigate gastric cancer associated biomarkers. In this research, gene expression data and DNA methylation data are integrated to analyze gastric cancer, and a feature selection approach is proposed to identify the possible biomarkers of gastric cancer. After the original data are pre-processed, the mutual information (MI) is applied to select some top genes. Then, fold change (FC) and T-test are adopted to identify differentially expressed genes (DEG). In particular, false discover rate (FDR) is introduced to revise p_value to further screen genes. For chosen genes, a deep neural network (DNN) model is utilized as the classifier to measure the quality of classification. The experimental results show that the approach can achieve superior performance in terms of accuracy and other metrics. Biological analysis for chosen genes further validates the effectiveness of the approach.

A comprehensive review of scaffolding methods in genome assembly.

In the field of genome assembly, scaffolding methods make it possible to obtain a more complete and contiguous reference genome, which is the cornerstone of genomic research. Scaffolding methods typically utilize the alignments between contigs and sequencing data (reads) to determine the orientation and order among contigs and to produce longer scaffolds, which are helpful for genomic downstream analysis. With the rapid development of high-throughput sequencing technologies, diverse types of reads have emerged over the past decade, especially in long-range sequencing, which have greatly enhanced the assembly quality of scaffolding methods. As the number of scaffolding methods increases, biology and bioinformatics researchers need to perform in-depth analyses of state-of-the-art scaffolding methods. In this article, we focus on the difficulties in scaffolding, the differences in characteristics among various kinds of reads, the methods by which current scaffolding methods address these difficulties, and future research opportunities. We hope this work will benefit the design of new scaffolding methods and the selection of appropriate scaffolding methods for specific biological studies.

LROD: An Overlap Detection Algorithm for Long Reads Based on k-mer Distribution.

Third-generation sequencing technologies can produce large numbers of long reads, which have been widely used in many fields. When using long reads for genome assembly, overlap detection between any pair of long reads is an important step. However, the sequencing error rate of third-generation sequencing technologies is very high, and obtaining accurate overlap detection results is still a challenging task. In this study, we present a long-read overlap detection (LROD) algorithm that can improve the accuracy of overlap detection results. To detect overlaps between two long reads, LROD first retains only the solid common k-mers between them. These k-mers can simplify the process of overlap detection. Second, LROD finds a chain (i.e., candidate overlap) that includes the consistent common k-mers. In this step, LROD proposes a two-stage strategy to evaluate whether two common k-mers are consistent. Finally, LROD uses a novel strategy to determine whether the candidate overlaps are true and to revise them. To verify the performance of LROD, three simulated and three real long-read datasets are used in the experiments. Compared with two other popular methods (MHAP and Minimap2), LROD can achieve good performance in terms of the F1-score, precision and recall. LROD is available from https://github.com/luojunwei/LROD.

Feature Selection for Microarray Data Classification Using Hybrid Information Gain and a Modified Binary Krill Herd Algorithm.

Due to the presence of irrelevant or redundant data in microarray datasets, capturing potential patterns accurately and directly via existing models is difficult. Feature selection (FS) has become a necessary strategy to identify and screen out the most relevant attributes. However, the high dimensionality of microarray datasets poses a serious challenge to most existing FS algorithms. For this purpose, we propose a novel feature selection strategy in this paper, called IG-MBKH. A pre-screening method of feature ranking which is based on information gain (IG) and an improved binary krill herd (MBKH) algorithm are integrated in this strategy. When searching for feature subsets using MBKH, a hyperbolic tangent function, an adaptive transfer factor, and a chaos memory weight factor are introduced to facilitate a better searching the possible feature subsets. The results indicates that the IG-MBKH algorithm can achieve improvement in convergence, the number of features and classification accuracy when compared to the BKH, MBKH, and several newest algorithms. Furthermore, we evaluate the impact of different classifiers on the performance of the strategy we propose.