High-precision segmentation of liver vessels is crucial for surgical planning and clinical diagnosis. Yet, it remains challenging due to intricate vascular structures and the low contrast inherent in CT images. We propose GASNet, an innovative liver vessel segmentation network that integrates progressive resolution-aware supervision with a learnable Gabor multi-filter module. GASNet generates hierarchical segmentation masks and applies adaptive supervision at intermediate layers to enhance multi-scale feature learning. Built upon a ConvNeXt backbone, the network integrates a Multi-Scale Feature Refiner and Dilated Convolutional modules to improve structural modeling capacity. We evaluate GASNet on two public datasets (LiVS and MSD) as well as aselfconstructed dataset, LVTSD. GASNet consistently outperforms state-of-the-art methods and demonstrates superior capability in distinguishing hepatic and portal veins on the revised MSD dataset, achieving a Dice score of 0.836 for hepatic veins and 0.820 for portal veins. The implementation is available at https://github.com/HappyBot516/GASNet.

Protein-protein interactions (PPIs) play a critical role in cellular functions, which are essential for maintaining the proper physiological state of organisms. Therefore, identifying PPI sites with high accuracy is crucial. Recently, graph neural networks (GNNs) have achieved significant progress in predicting PPI sites, but there is still potential for further enhancement. In this study, we introduce GTE-PPIS, an innovative PPI site predictor that utilizes two components: a graph transformer and an equivariant GNN, to collaboratively extract features. These extracted features are subsequently processed through a multilayer perceptron to generate the final predictions. Our experimental results show that GTE-PPIS consistently outperforms existing methods on multiple evaluation metrics across benchmark datasets, strongly supporting the effectiveness of our approach.

Abstract Deep potential (DP) scheme has increased the simulation temporal and spatial scales while maintaining the ab initio accuracy of the molecular dynamics. DeePMD-kit is an outstanding application that implements DP scheme efficiently. However, current performance model cannot accurately measure the resource utilization of DeePMD-kit operators and predict the execution time. We introduce DP-perf, an interpretable performance model for DeePMD-kit. DP-perf can accurately measure the resource utilization of the individual DeePMD-kit operators, communication pattern, and the overall application by exploiting physical system properties and machine configurations. It can be easily applied to mainstream supercomputers including Tianhe-3F, the new Sunway, Fugaku, and Summit. With DP-perf, users can select the optimal machine and decide the corresponding configuration for various purposes (e.g., lower cost, less time) without real runs. Evaluation of four top supercomputers shows that DP-perf can fit overall execution time with a low mean absolute percentage error of 5.7 %/8.1%/14.3%/13.1% on Tianhe-3F/new Sunway/Fugaku/Summit. On the prediction scenario, DP-perf can predict the total execution time with a mean absolute percentage error of less than 20%.

Physical phenomena such as bond breaking and phase transitions require molecular dynamics (MD) withab initioaccuracy, involving up to billions of atoms and over nanosecond timescales. Previous state-of-the-art work has demonstrated that neural network molecular dynamics (NNMD) like deep potential molecular dynamics (DeePMD), can successfully extend the temporal and spatial scales of MD withab initioaccuracy on both ARM and GPU platforms. However, the DeePMD-kit package is currently unable to fully exploit the computational potential of the new Sunway supercomputer due to its unique many-core architecture, memory hierarchy, and low precision capability. In this paper, we re-design the DeePMD-kit to harness the massive computing power of the new Sunway, enabling the MD with over ten billion atoms. We first design a large-scale parallelization scheme to exploit the massive parallelism of the new Sunway. Then we devise specialized optimizations for the time-consuming operators. Finally, we design a novel mixed precision method for DeePMD-kit customized operators to leverage the low precision computing power of the new Sunway. The optimized DeePMD-kit achieves 67.6 / 56.5$\boldsymbol{\times}$speedup for water / copper systems on the new Sunway. Meanwhile, it can perform 29 billion atoms simulation for the water system on 35 million cores (i.e., 90,000 computing nodes, around 84% of the whole supercomputer) with a peak performance of 57.1 PFLOPs, which is 7.9$\boldsymbol{\times}$bigger and 1.2$\boldsymbol{\times}$faster than state-of-the-art results. This paves the way for investigating more realistic scenarios, such as studying the mechanical properties of metals, semiconductor devices, batteries, and other materials and physical systems.

Improved cancer genomic diagnosis and prognosis are vital to accurate medical therapy. Deep learning methods offered an end-to-end solution to enhance the precision of analysis. With the fast pace of pre-trained Transformer models, it remains uncertain whether some novel approaches such as the sparsely gated mixture of expert (MOE) and self-attention mechanisms can further improve the precision of cancer prognosis and classification. In this paper, we introduce a novel sparsely gated cancer diagnosis and prognosis framework called Gene-MOE exploiting the potential of the MOE layers and the proposed mixture of attention expert (MOAE) layers to enhance the analysis accuracy. Additionally, we address overfitting challenges by integrating pan-cancer information from 33 distinct cancer types through pre-training. For survival analysis, Gene-MOE achieves the best Concordance Index compared with state-of-the-art models on 12 of 14 cancer types. For cancer classification, the total accuracy of the classification model for 33 cancer classifications reached 95.8%, representing the best performance compared to state-of-the-art models. For cancer subtyping, Gene-MOE achieves the best result on at least one metric of the log10 P-values and the number of significant clinical on seven of nine cancers. These results indicate that Gene-MOE holds strong potential for these downstream tasks.

The annotation of cell types based on single-cell RNA sequencing (scRNA-seq) data is a critical downstream task in single-cell analysis, with significant implications for a deeper understanding of biological processes. Most analytical methods cluster cells by unsupervised clustering, which requires manual annotation for cell type determination. This procedure is time-overwhelming and non-repeatable. To accommodate the exponential growth of sequencing cells, reduce the impact of data bias, and integrate large-scale datasets for further improvement of type annotation accuracy, we proposed scSwinTNet. It is a pre-trained tool for annotating cell types in scRNA-seq data, which uses self-attention based on shifted windows and enables intelligent information extraction from gene data. We demonstrated the effectiveness and robustness of scSwinTNet by using 399 760 cells from human and mouse tissues. To the best of our knowledge, scSwinTNet is the first model to annotate cell types in scRNA-seq data using a pre-trained shifted window attention-based model. It does not require a priori knowledge and accurately annotates cell types without manual annotation.

The identification of protein-protein interaction sites (PPIS) plays a crucial role in understanding the mechanisms of biological processes. Traditional biological experimental methods for PPIS prediction are both expensive and time-consuming, developing computational methods can effectively reduce costs. However, existing approaches often focus on single-scale features and pay little attention to spatial neighborhood features, leading to unsatisfactory prediction performance. To address these challenges, we propose a dual-branch PPIS predictor (AEG-PPIS) based on augmented graph attention network (AGAT) and E(n) equivariant graph neural network (EGNN). AEG-PPIS extracts global features through EGNN, ensuring rotational and translational invariance of the protein graph. For local feature extraction, it employs an enhanced Augmented Graph Attention Network, which integrates initial node features, previous layer outputs, and features extracted by GraphSAGE using residual connections and identity mapping. Our model realizes the modeling of multi-scale information. Comparative experimental results show that the performance of AEG-PPIS is better than that of state-of-the-art methods. Ablation experiments and case studies demonstrate the effectiveness of AEG-PPIS in predicting PPIS.

Gene expression is a pivotal biological process within organisms, and in recent years, the prediction of gene expression levels has garnered increasing attention due to its vast potential in clinical applications. Predicting gene expression levels is a complex problem as gene expression is influenced by multiple factors, including but not limited to gene sequences, epigenetic modifications, transcription factor binding, and micro-environmental conditions. This paper proposes a model named Multimodal Expression, based on the Transformer architecture, which integrates various data types. The model can extract effective features from gene promoter sequences and combine pre-transcriptional and post-transcriptional regulatory information to predict gene expression levels. Experimental results demonstrate that our model can extract more effective information from promoter sequences, and the attention mechanism in the Transformer can integrate multiple data types to jointly predict gene expression levels. Compared to previous methods, our model’s R2values improved by 7.05%, 8.9%, and 1.91% when using gene sequence data alone, gene sequence data combined with mRNA half-life data, and gene sequence data combined with mRNA half-life data and transcription factor data, respectively.

Efficient and accurate identification of drug-target affinity (DTA) is crucial in the virtual screening stage of drug discovery and the reuse of existing drugs. Deep neural networks are becoming popular tools for providing fast and accurate binding affinity prediction. MGraphDTA is one of the best performing models for DTA prediction, utilizing a super-deep graph neural network to extract features, demonstrating good generalization and interpretation capabilities. However, the complex deep neural networks come with a higher computational cost of model inference. Meanwhile, due to the expansion of drug databases, rapid completion of large-scale virtual screening still presents challenges. In this work, we propose an efficient binding affinity prediction method called DCUI-MGraphDTA through a series of optimizations on MGraphDTA to achieve large-scale virtual screening for millions of molecules on the Deep Computing Unit (DCU) cluster. The optimized DCUI-MGraphDTA model achieves a speedup of approximately 5.83 times on a single DCU compared to the original MGraphDTA. As the number of DCUs expands from 1 to 32, the total time for the model to process 1.28 million data significantly decreases, achieving a speedup of approximately 11.9 times.

Depression is a prevalent global mental health issue that significantly impacts both individuals and societies. Current methods often overlook the heterogeneity of depressive symptoms and fail to fully leverage available data, underscoring the need for more advanced computational methods DTI prediction. This paper introduces KGM4DTI, a multi-semantic, multi-view, and multi-level meta-path Transformer framework that integrates Transformer and GNN. KGM4DTI effectively captures both global semantic and local structural information between interaction pairs within a depression knowledge graph (DepKG), which comprises 2,258 drugs, 3,360 proteins, 6 subtypes of depression, 196 phenotypes, and 238 pathways. Comparative experiments have demonstrated that KGM4DTI significantly outperforms existing state-of-the-art methods, achieving an AUROC of 99.72 and an AUPR of 99.76. The robustness of framework is further confirmed through extensive ablation studies, which highlight the crucial role of integrating both global and local information. These results underscore the potential of KGM4DTI in accurately predicting DTI. This approach has promising implications for drug discovery, particularly in developing effective treatments for depression, and could extend its applicability to broader mental health disorders. The data and code are available at https://github.com/Tianxyuu/KGM4DTI/.

Fourier Neural Operator (FNO) has been proven to be a universal and effective deep learning framework capable of achieving remarkable accuracy on Partial Differential Equation (PDE) solution problem. However, certain key components of emerging FNO-based models cannot leverage hardware potential, which makes it difficult to apply in high resolution and high realtime demand scenario. This paper presents a high optimized model called Speed Galerkin Transformer, including multilevel parallel SliceK-SplitK-ReduceK strategy for batched skinny matrix multiplication, memory layout optimization for QKV matrices and positional encodings and multi-head layer normalization fusion, as well as batched transposition optimization with strided scattering and gathering in 2D FNO, and these strategies can achieve $10.29 \mathrm{x}, 4.41 \mathrm{x}$ and 2.38 x speedup respectively under specific configuration. When solving the Darcy Flow equation at 512x512 resolution, the Speed Galerkin Transformer model can achieve about 1.72 x speedup, and achieve more than $\mathbf{9 0 \%}$ parallel efficiency on 8 GPUs.

The identification of transcription factor binding sites (TFBSs) is crucial for understanding the regulatory mechanisms of gene expression, which contributes to unraveling cellular functions and disease development. Currently, the most common approach involves the use of deep learning techniques to predict TFBSs by combining sequence and shape features. Although significant progress has been made with these methods, the integration of local features extracted from DNA sequences and shapes with global features has not yet reached a sufficient level, and there is still significant room for improvement in the accuracy of prediction results. In this paper, we propose a novel framework based on convolution and attention mechanisms, referred to as TBCA, which combines DNA sequence information and shape information for predicting transcription factor binding sites. In this work, we employ a two-layer convolutional neural network (CNNs) and self-attention mechanism to extract complex sequence features from DNA. What's more, we utilize a Fourier-transform-enhanced multi-head attention along with channel attention to extract high-order shape features of DNA. Finally, these high-order sequence and shape features are integrated into the channel dimension to achieve accurate TFBSs prediction. Our research results demonstrate that TBCA exhibits superior predictive performance in 165 validated ChIP-seq datasets. Furthermore, the employed attention mechanisms can automatically learn important features at different positions and scales, enhancing the accuracy and robustness of feature representation. We also conduct an in-depth analysis of the contributions of five different shapes to site prediction, revealing that shape features can enhance the prediction of transcription factor DNA binding.

Predicting protein function from sequences through machine learning can improve the understanding of novel proteins and biological mechanisms. Existing methods mainly rely on one-dimensional convolution or natural language processing (NLP) techniques to extract features from sequences, but they suffer from limited predictive performance. To address this challenge, we propose MulAxialGO, a new method that leverages multi-modal feature fusion to improve prediction accuracy. MulAxialGO integrates the prior features of a large-scale pre-trained protein language model and the posterior features of dynamic embedding coding and sequence homology. In addition, MulAxialGO employs a comprehensive image feature encoder to extract features from sequences, providing a novel perspective for protein function prediction. MulAxialGO is tested on two benchmark datasets and achieves state-of-the-art results. On the 2016 dataset, MulAxialGO significantly outperforms DeepGOPlus, improving molecular function by 4.5 points, biological process by 2.4 points and cellular component by 1.6 points for the AUPR metric. Similarly, on the NetGO dataset, MulAxialGO outperforms the state-of-the-art NetGO2.0, improving Fmax by 1.1 points for biological process and 2.3 points for cellular component.

Liver cancer is one of the most common malignant diseases worldwide. Segmentation and reconstruction of liver tumors and vessels in CT images can provide convenience for physicians in preoperative planning and surgical intervention. In this paper, we introduced a TransFusionNet framework, which consists of a semantic feature extraction module, a local spatial feature extraction module, an edge feature extraction module, and a multi-scale feature fusion module to achieve fine-grained segmentation of liver tumors and vessels. In addition, we applied the transfer learning approach to pre-train using public datasets and then fine-tune the model to further improve the fitting effect. Furthermore, we proposed an intelligent quantization scheme to compress the model weights and achieved high performance inference on JetsonTX2. The TransFusionNet framework achieved mean IoU of 0.854 in vessel segmentation task, and achieved mean IoU of 0.927 in liver tumor segmentation task. When profiling the Computational Performance of the quantized inference, our quantized model achieved 4TFLOPs on Node with NVIDIA RTX3090 and 132GFLOPs on JetsonTX2. This unprecedented segmentation effect solves the accuracy and performance bottleneck of automated segmentation to a certain extent.

In the data space, time-series analysis has emerged in many fields, including biology, healthcare, and numerous large-scale scientific facilities like astronomy, climate science, particle physics, and genomics. Clustering is one of the most critical methods in time-series analysis. So far, the state-of-art time series clustering algorithm k-Shape has been widely used not only because of its high accuracy, but also because of its relatively low computation cost. However, due to the high heterogeneity of time series data, it can not be simply regarded as a high-dimensional vector. Two time series often need some alignment method in similarity comparison. The alignment between sequences is often a time-consuming process. For example, when using dynamic time warping as a sequence alignment algorithm and if the length of time series is greater than 1,000, a single iteration in the clustering process may take hundreds to tens of thousands of seconds, while the entire clustering cycle often requires dozens of iterations. In this article, we propose a set of novel parallel strategies suitable for GPU's computation model, called Times-C, which is an abbreviation for Time Series Clustering. We define three stages in the analysis process: aggregation, centroid, and class assignment. Times-C includes efficient parallel algorithms and corresponding implementations for these three stages. Overall, the experimental results show that the Times-C algorithm exhibits a performance improvement of one to two orders of magnitude compared to the multi-core CPU version of k-Shape. Furthermore, compared to the GPU version of the k-Shape algorithm, the Times-C algorithm achieves a maximum acceleration of up to 345 times.

With the development of three-generation genome sequencing technology, targeted sequencing has become a fundamental need in virus detection, metagenomics expansion, and human polymorphisms detection. However, previous studies on CPU/GPU platforms have much lower throughput than Nanopore sequencers, and hardware designs have high design cost and flexibility issues. In this paper, we propose Crescent, a GPU-based targeted sequence selector. Specifically, on the algorithm level, we modify the popular-used algorithm without significantly decreasing the accuracy and then propose four key observations. On the architecture level, based on four key observations, we propose two-stage parallelism mechanisms (i.e., inter-task and intra-task) and other optimization strategies (i.e., data reuse and redundant computation elimination) to accelerate our modified algorithm. Our experimental results show that the throughput of Crescent on NVIDIA A100 is $10.84 \times$ and $2.30 \times$ higher than a state-of-the-art 16-thread and 72-thread CPU baseline, respectively.

Booming data volume has become an important challenge for data center storage and bandwidth resources. Consequently, fast and efficient compression architecture is becoming the most fundamental design in data centers. However, the compression ratio (CR) and compression throughput are often difficult to achieve at the same time on existing computing platforms. DEFLATE is a widely used compression format in data centers, which is an ideal case for hardware acceleration. Unfortunately, Deflate has an inherent connection among its special memory access pattern, which limits a higher throughput.

The properties of the drug may be altered by the combination, which may cause unexpected drug-drug interactions (DDIs). Prediction of DDIs provides combination strategies of drugs for systematic and effective treatment. In most of deep learning-based methods for predicting DDI, encoded information about the drugs is insufficient in some extent, which limits the performances of DDIs prediction. In this work, we propose a novel attention-mechanism-based multidimensional feature encoder for DDIs prediction, namely attention-based multidimensional feature encoder (AMDE). Specifically, in AMDE, we encode drug features from multiple dimensions, including information from both Simplified Molecular-Input Line-Entry System sequence and atomic graph of the drug. Data experiments are conducted on DDI data set selected from Drugbank, involving a total of 34 282 DDI relationships with 17 141 positive DDI samples and 17 141 negative samples. Experimental results show that our AMDE performs better than some state-of-the-art baseline methods, including Random Forest, One-Dimension Convolutional Neural Networks, DeepDrug, Long Short-Term Memory, Seq2seq, Deepconv, DeepDDI, Graph Attention Networks and Knowledge Graph Neural Networks. In practice, we select a set of 150 drugs with 3723 DDIs, which are never appeared in training, validation and test sets. AMDE performs well in DDIs prediction task, with AUROC and AUPRC 0.981 and 0.975. As well, we use Torasemide (DB00214) as an example and predict the most likely drug to interact with it. The top 15 scores all have been reported with clear interactions in literatures.

Multi-drug combinations for the treatment of complex diseases are gradually becoming an important treatment, and this type of treatment can take advantage of the synergistic effects among drugs. However, drug-drug interactions (DDIs) are not just all beneficial. Accurate and rapid identifications of the DDIs are essential to enhance the effectiveness of combination therapy and avoid unintended side effects. Traditional DDIs prediction methods use only drug sequence information or drug graph information, which ignores information about the position of atoms and edges in the spatial structure. In this paper, we propose Molormer, a method based on a lightweight attention mechanism for DDIs prediction. Molormer takes the two-dimension (2D) structures of drugs as input and encodes the molecular graph with spatial information. Besides, Molormer uses lightweight-based attention mechanism and self-attention distilling to process spatially the encoded molecular graph, which not only retains the multi-headed attention mechanism but also reduces the computational and storage costs. Finally, we use the Siamese network architecture to serve as the architecture of Molormer, which can make full use of the limited data to train the model for better performance and also limit the differences to some extent between networks dealing with drug features. Experiments show that our proposed method outperforms state-of-the-art methods in Accuracy, Precision, Recall and F1 on multi-label DDIs dataset. In the case study section, we used Molormer to make predictions of new interactions for the drugs Aliskiren, Selexipag and Vorapaxar and validated parts of the predictions. Code and models are available at https://github.com/IsXudongZhang/Molormer.

Human papillomavirus (HPV) infection is linked to several diseases, the most prominent of which are cervical cancer and genital condyloma acuminatum. PI3K-Akt-mTOR signaling pathway is one of the most important signaling pathway in the regulation of proliferation, differentiation and apoptosis of human cervical cancer cells, and this pathway has the potential to become a novel target for the development of cervical cancer therapeutics. Previous studies have suggested that drug therapy can modulate PI3K-AKT-mTOR pathway to reduce HPV viral load effectively through autophagy and apoptosis. Therefore, in our study, in order to find drugs that can regulate this pathway, we collected more than 20 proteins related to this pathway and used deep learning models to predict protein-ligand binding affinity, finally listed the drug molecules with the highest predicted affinity score for each protein molecule.

Traditional Chinese medicines (TCMs) have been used to treat diseases for thousands of years. The application of traditional Chinese medicine provides new ideas for the treatment of cancer and other intractable diseases. Phosphoinositide-3kinase (PI3K) pathway is an important way to regulate tumor cells, such as cervical cancer. Deep learning provides a powerful application in calculating interactions between drugs and targets. In this study, we try to use the method of deep learning to reposition molecules of TCMs and 21 targets on PI3K pathway, and predict the TCMs that can regulate PI3K pathway, so as to achieve the purpose of cancer treatment. A deep convolutional neural network (DCNN) is constructed and trained on KIBA dataset. The accuracy of predicting the binding affinity of drug-target pairs is 85.3%. DCNN ranked 433 molecules of 35 TCMs with 21 PI3K pathway target proteins. We find that Gancao and Huangqin have strong binding affinity with more than half of PI3K pathway targets. Meanwhile, Renshen, Zhizi, Mahuang, etc. are also effective on multiple targets.

Ovarian cancer is a malignant tumor that poses a serious threat to women’s lives. Computer-aided diagnosis (CAD) systems can classify the type of ovarian tumors, but few of them can provide exactly the location information of ovarian cancer cells. Recently, deep learning technology becomes hot for automatic detection of cancer cells, particularly for detecting their locations. In this work, we propose a novel end-to-end network YOLO-OC (Ovarian cancer) model, which can extract the characteristics of ovarian cancer more efficiently. In our method, deformable convolution is used to enhance the model’s ability to learn geometric deformation in space. Squeeze-and-Excitation (SE) module is proposed to automatically learn the importance of different channel features. Data experiments are conducted on datasets collected from The Affiliated Hospital of Qingdao University Medical College, China. Experimental results show that our YOLO-OC model achieves 91.83%, 85.66% and 73.82% on mean average precision [email protected], [email protected] and mAP@[.5,.95], respectively, which performs better than Faster R-CNN, SSD and RetinaNet on both accuracy and efficiency.