Prediction of the tetramer protein complex interaction based on CNN and SVM

Identifying the DNA methylation preference of transcription factors using ProtBERT and SVM

Transcription factors (TFs) can affect gene expression by binding to certain specific DNA sequences. This binding process of TFs may be modulated by DNA methylation. A subset of TFs that serve as methylation readers preferentially binds to certain methylated DNA and is defined as TFPM. The identification of TFPMs enhances our understanding of DNA methylation's role in gene regulation. However, their experimental identification is resource-demanding. In this study, we propose a novel two-step computational approach to classify TFs and TFPMs. First, we employed a fine-tuned ProtBERT model to differentiate between the classes of TFs and non-TFs. Second, we combined the Reduced Amino Acid Category (RAAC) with K-mer and SVM to predict the potential of TFs to bind to methylated DNA. Comparative experiments demonstrate that our proposed methods outperform all existing approaches and emphasize the efficiency of our computational framework in classifying TFs and TFPMs. Cross-species validation on an independent mouse dataset further demonstrates the generalizability of our proposed framework In addition, we conducted predictions on all human transcription factors and found that most of the top 20 proteins belong to the Krueppel C2H2-type Zinc-finger family. So far, some studies have demonstrated a partial correlation between this family and DNA methylation and confirmed the preference of some of its members, thereby showing the robustness of our approach.

Prediction of the Trimer Protein Interface Residue Pair by CNN-GRU Model Based on Multi-Feature Map

Most life activities of organisms are realized through protein-protein interactions, and these interactions are mainly achieved through residue-residue contact between monomer proteins. Consequently, studying residue-residue contact at the protein interaction interface can contribute to a deeper understanding of the protein-protein interaction mechanism. In this paper, we focus on the research of the trimer protein interface residue pair. Firstly, we utilize the amino acid k-interval product factor descriptor (AAIPF(k)) to integrate the positional information and physicochemical properties of amino acids, combined with the electric properties and geometric shape features of residues, to construct an 8 × 16 multi-feature map. This multi-feature map represents a sample composed of two residues on a trimer protein. Secondly, we construct a CNN-GRU deep learning framework to predict the trimer protein interface residue pair. The results show that when each dimer protein provides 10 prediction results and two protein-protein interaction interfaces of a trimer protein needed to be accurately predicted, the accuracy of our proposed method is 60%. When each dimer protein provides 10 prediction results and one protein-protein interaction interface of a trimer protein needs to be accurately predicted, the accuracy of our proposed method is 93%. Our results can provide experimental researchers with a limited yet precise dataset containing correct trimer protein interface residue pairs, which is of great significance in guiding the experimental resolution of the trimer protein three-dimensional structure. Furthermore, compared to other computational methods, our proposed approach exhibits superior performance in predicting residue-residue contact at the trimer protein interface.

TCellR2Vec: efficient feature selection for TCR sequences for cancer classification

Cancer remains one of the leading causes of death globally. New immunotherapies that harness the patient's immune system to fight cancer show promise, but their development requires analyzing the diversity of immune cells called T-cells. T-cells have receptors that recognize and bind to cancer cells. Sequencing these T-cell receptors allows to provide insights into their immune response, but extracting useful information is challenging. In this study, we propose a new computational method, TCellR2Vec, to select key features from T-cell receptor sequences for classifying different cancer types. We extracted features like amino acid composition, charge, and diversity measures and combined them with other sequence embedding techniques. For our experiments, we used a dataset of over 50,000 T-cell receptor sequences from five cancer types, which showed that TCellR2Vec improved classification accuracy and efficiency over baseline methods. These results demonstrate TCellR2Vec's ability to capture informative aspects of complex T-cell receptor sequences. By improving computational analysis of the immune response, TCellR2Vec could aid the development of personalized immunotherapies tailored to each patient's T-cells. This has important implications for creating more effective cancer treatments based on the individual's immune system.

User emotion recognition and indoor space interaction design: a CNN model optimized by multimodal weighted networks

In interior interaction design, achieving intelligent user-interior interaction is contingent upon understanding the user's emotional responses. Precise identification of the user's visual emotions holds paramount importance. Current visual emotion recognition methods rely solely on singular features, predominantly facial expressions, resulting in inadequate coverage of visual characteristics and low recognition rates. This study introduces a deep learning-based multimodal weighting network model to address this challenge. The model initiates with a convolutional attention module, employing a self-attention mechanism within a convolutional neural network (CNN). As a result, the multimodal weighting network model is integrated to optimize weights during training. Finally, a weight network classifier is derived from these optimized weights to facilitate visual emotion recognition. Experimental outcomes reveal a 77.057% correctness rate and a 74.75% accuracy rate in visual emotion recognition. Comparative analysis against existing models demonstrates the superiority of the multimodal weight network model, showcasing its potential to enhance human-centric and intelligent indoor interaction design.

Deep learning in modeling protein complex structures: From contact prediction to end-to-end approaches

Protein-protein interactions play crucial roles in many biological processes. Traditionally, protein complex structures are normally built by protein-protein docking. With the rapid development of artificial intelligence and its great success in monomer protein structure prediction, deep learning has widely been applied to modeling protein-protein complex structures through inter-protein contact prediction and end-to-end approaches in the past few years. This article reviews the recent advances of deep-learning-based approaches in modeling protein-protein complex structures as well as their advantages and limitations. Challenges and possible future directions are also briefly discussed in applying deep learning for the prediction of protein complex structures.