RNA modification, which is evolutionarily conserved, is crucial for modulating various biological functions and disease pathogenesis. High resolution transcriptome-wide mapping of RNA modifications has facilitated both data resources and computational prediction of RNA modification. While these prediction algorithms are promising, they are limited in interpretability or generalizability, or the capacity for discovering novel post-transcriptional regulations. Here, we present NetRNApan, a deep learning framework for RNA modification site prediction, motif discovery and trans-regulatory factor identification. Using m5U profiles generated by FICC-seq and miCLIP-seq technologies and single-base resolution m6A sites from multiple experiments as cases, we demonstrated the accuracy of NetRNApan with more efficient and interpretive feature representations. For m5U modification, we uncovered five representative clusters with consensus motifs that may be essential by decoding the informative characteristics detected by NetRNApan. Furthermore, NetRNApan revealed interesting trans-regulatory factors and provided a protein-binding perspective for investigating the function of RNA modifications. Specifically, we discovered 21 potential functional RNA-binding proteins (RBPs) whose binding sites were significantly linked to the extracted top-scoring motifs for m5U modification. Two examples are ANKHD1 and RBM4 with potential regulatory function of m5U modifications. Meanwhile, the analysis of convolution layer parameters within the model offers valuable insights into the regulation of m6A in humans. Collectively, NetRNApan demonstrated high accuracy, interpretability and generalizability for study of RNA modification and mRNA regulation. NetRNApan is freely available at https://github.com/bsml320/NetRNApan.

Congestion control is one of the fundamental techniques for ensuring the stability and reliability of high-performance network applications. During the research on congestion control strategies for existing high-speed RDMA networks, it was observed that the current strategies are primarily applied in structured and stable data center networks, which are difficult to generalize to network environments with slightly poorer performance and packet loss. In this paper, we implement and analyze a receiver-driven end-to-end congestion control scheme called Shaped Quota for complex networks. Shaped Quota achieves rate control by actively evaluating the network congestion status of each flow at the receiver and providing feedback to the corresponding sender. Building upon this research, this paper further refines the design of the Shaped Quota and implements the algorithm for the first time in a simulated environment, followed by comprehensive testing. The experiments demonstrate that in lossless networks, Shaped Quota can provide superior transmission efficiency compared to existing congestion control schemes. Under lossy network conditions, Shaped Quota improves the reliability and stability of network transmission. Compared to DCQCN, Timely, and HPCC, Shaped Quota achieves an average reduction of 45.26% in Flow Completion Time (FCT) in lossless networks. In lossy networks, the flow completion rate of Shaped Quota is significantly higher, reaching 29.33 times and 29.24 times that of DCQCN and Timely, respectively. Compared to HPCC, which exhibits better queue length control, Shaped Quota reduces the queue length by more than 40%.

Survival prediction based on histopathological whole slide images (WSIs) is of great significance for risk-benefit assessment and clinical decision. However, complex microenvironments and heterogeneous tissue structures in WSIs bring challenges to learning informative prognosis-related representations. Additionally, previous studies mainly focus on modeling using mono-scale WSIs, which commonly ignore useful subtle differences existed in multi-zoom WSIs. To this end, we propose a deep multi-dictionary learning framework for cancer survival prediction with multi-zoom histopathological WSIs. The framework can recognize and learn discriminative clusters (i.e., microenvironments) based on multi-scale deep representations for survival analysis. Specifically, we learn multi-scale features based on multi-zoom tiles from WSIs via stacked deep autoencoders network followed by grouping different microenvironments by cluster algorithm. Based on multi-scale deep features of clusters, a multi-dictionary learning method with a post-pruning strategy is devised to learn discriminative representations from selected prognosis-related clusters in a task-driven manner. Finally, a survival model (i.e., EN-Cox) is constructed to estimate the risk index of an individual patient. The proposed model is evaluated on three datasets derived from The Cancer Genome Atlas (TCGA), and the experimental results demonstrate that it outperforms several state-of-the-art survival analysis approaches.

The use of RDMA in datacenter networks is gradually increasing. However, due to the relatively simple flow control and congestion control mechanisms in RDMA networks, they lack the ability to adapt to various network loads and environments. Therefore, congestion can easily occur under high load, which severely affects network performance. Recently, a new congestion control algorithm called Shaped-Quota was proposed, and the main ideas of the algorithm include receiver-driven, traffic shaping, and bandwidth reservation. In this paper, we analyze the short message limit N and bandwidth reservation ratio R mentioned in Shaped-Quota, provide a definition formula, and conduct experiments, which can help the Shaped-Quota algorithm better apply to various network scales, topologies, and traffic models.

The multi-instance learning (MIL) has advanced cancer prognosis analysis with whole slide images (WSIs). However, current MIL methods for WSI analysis still confront unique challenges. Previous methods typically generate instance representations via a pre-trained model or a model trained by the instances with bag-level annotations, which, however, may not generalize well to the downstream task due to the introduction of excessive label noises and the lack of fine-grained information across multi-magnification WSIs. Additionally, existing methods generally aggregate instance representations as bag ones for prognosis prediction and have no consideration of intra-bag redundancy and inter-bag discrimination. To address these issues, we propose a dual-curriculum contrastive MIL method for cancer prognosis analysis with WSIs. The proposed method consists of two curriculums, i.e., saliency-guided weakly-supervised instance encoding with cross-scale tiles and contrastive-enhanced soft-bag prognosis inference. Extensive experiments on three public datasets demonstrate that our method outperforms state-of-the-art methods in this field. The code is available at https://github.com/YuZhang-SMU/Cancer-Prognosis-Analysis/tree/main/DC_MIL%20Code.

The integrative analysis of complementary phenotype information contained in multi-modality data (e.g., histopathological images and genomic data) has advanced the prognostic evaluation of cancers. However, multi-modality based prognosis analysis confronts two challenges: (1) how to explore underlying relations inherent in different modalities data for learning compact and discriminative multi-modality representations; (2) how to take full consideration of incomplete multi-modality data for constructing accurate and robust prognostic model, since a host of complete multi-modality data are not always available. Additionally, many existing multi-modality based prognostic methods commonly ignore relevant clinical variables (e.g., grade and stage), which, however, may provide supplemental information to promote the performance of model. In this paper, we propose a relation-aware shared representation learning method for prognosis analysis of cancers, which makes full use of clinical information and incomplete multi-modality data. The proposed method learns multi-modal shared space tailored for prognostic model via a dual mapping. Within the shared space, it equips with relational regularizers to explore the potential relations (i.e., feature-label and feature-feature relations) among multi-modality data for inducing discriminatory representations and simultaneously obtaining extra sparsity for alleviating overfitting. Moreover, it regresses and incorporates multiple auxiliary clinical attributes with dynamic coefficients to meliorate performance. Furthermore, in training stage, a partial mapping strategy is employed to extend and train a more reliable model with incomplete multi-modality data. We have evaluated our method on three public datasets derived from The Cancer Genome Atlas (TCGA) project, and the experimental results demonstrate the superior performance of the proposed method.