Search

Rapidly developing single-cell sequencing analyses produce more comprehensive profiles of the genomic, transcriptomic, and epigenomic heterogeneity of tumor subpopulations than do traditional bulk sequencing analyses. Moreover, single-cell techniques allow the response of a tumor to drug exposure to be more thoroughly investigated. Deep learning (DL) models have successfully extracted features from complex bulk sequence data to predict drug responses. We review recent innovations in single-cell technologies and DL-based approaches related to drug sensitivity predictions. We believe that, by using insights from bulk sequence data, deep transfer learning (DTL) can facilitate the use of single-cell data for training superior DL-based drug prediction models. A comprehensive understanding of heterogeneous tumor subpopulations will benefit drug sensitivity prediction and combination drug treatment design. Deep learning models are powerful and extensively used in drug sensitivity prediction and in inferring drug–target interactions. Single-cell sequencing techniques offer precise and accurate profiling of tumor subpopulations and reveal subtle differences in their response to drug treatments. Applying deep transfer learning to predict drug sensitivity allows us to not only take advantage of prior knowledge obtained from massive bulk sequencing data but also utilize the heterogeneous landscapes generated by single-cell sequencing techniques. The integration of single-cell multi-omic data for drug sensitivity prediction using transfer learning methods poses a special challenge. [ABSTRACT FROM AUTHOR]

Accurately predicting peptide secondary structures remains a challenging task due to the lack of discriminative information in short peptides. In this study, PHAT is proposed, a deep hypergraph learning framework for the prediction of peptide secondary structures and the exploration of downstream tasks. The framework includes a novel interpretable deep hypergraph multi‐head attention network that uses residue‐based reasoning for structure prediction. The algorithm can incorporate sequential semantic information from large‐scale biological corpus and structural semantic information from multi‐scale structural segmentation, leading to better accuracy and interpretability even with extremely short peptides. The interpretable models are able to highlight the reasoning of structural feature representations and the classification of secondary substructures. The importance of secondary structures in peptide tertiary structure reconstruction and downstream functional analysis is further demonstrated, highlighting the versatility of our models. To facilitate the use of the model, an online server is established which is accessible via http://inner.wei‐group.net/PHAT/. The work is expected to assist in the design of functional peptides and contribute to the advancement of structural biology research. [ABSTRACT FROM AUTHOR]

• A novel method (GcForest-PPI) to predict protein–protein interactions. • The PseAAC, AD, MMI, CTD, AAC-PSSM and DPC-PSSM are fused to extract feature information. • The elastic net is employed to eliminate redundant and irrelevant features. • We firstly use deep forest as classifier to predict PPIs via layer-by-layer processing of raw features. • GcForest-PPI model has good generalization ability on cross-species datasets and PPIs network. Prediction of protein–protein interactions (PPIs) helps to grasp molecular roots of disease. However, web-lab experiments to predict PPIs are limited and costly. Using machine-learning-based frameworks can not only automatically identify PPIs, but also provide new ideas for drug research and development from a promising alternative. We present a novel deep-forest-based method for PPIs prediction. Firstly, pseudo amino acid composition (PAAC), autocorrelation descriptor (Auto), multivariate mutual information (MMI), composition-transition-distribution (CTD), amino acid composition position-specific scoring matrix (AAC-PSSM), and dipeptide composition PSSM (DPC-PSSM) are adopted to extract and construct the pattern of PPIs. Secondly, elastic net is utilized to optimize the initial feature vectors and boost the predictive performance. Finally, we ensemble XGBoost, random forest, and extremely randomized trees to construct deep forest model via cascade architecture for PPIs prediction (GcForest-PPI). Benchmark experiments reveal that the proposed approach outperforms other state-of-the-art predictors on Saccharomyces cerevisiae and Helicobacter pylori. We also apply GcForest-PPI on independent test sets, CD9-core network, crossover network, and cancer-specific network. The evaluation shows that GcForest-PPI can boost the prediction accuracy, complement experiments and improve drug discovery. [ABSTRACT FROM AUTHOR]

Lysine malonylation is one of the important post-translational modification (PTM) of proteins. Malonylated proteins can affect various cell functions of eukaryotes and prokaryotes, which play an important role in metabolic pathways, mitochondrial functions, fatty acid oxidation and other life activities. However, accurate identification of the malonylation sites is the key to further understand the molecular mechanism of malonylation. Currently, experimental identification is still a challenging task, which usually requires a large amount of laboratory work and considerable cost. Regarding this situation, there is an urgent need to establish useful calculation methods and develop effective predictors. We propose a new deep learning network model called DeepMal. Firstly, features are extracted by enhanced amino acid composition (EAAC), enhanced grouped amino acid composition (EGAAC), dipeptide deviation from the expected mean (DDE), K nearest neighbors (KNN) and BLOSUM62 matrix. Secondly, the linear convolutional neural network is used to extract the specific features of malonylation sites, select the relevant features and reduce the feature dimension through maximum pooling. Finally, malonylation sites and non-malonylation sites are classified by a multilayer neural network. And an independent dataset is used to assess the predictive ability of the model DeepMal. On the independent datasets Escherichia coli (E. coli), Homo sapiens (H. sapiens), Mus musculus (M. musculus), the AUC values are 0.974, 0.956 and 0.944, and the accuracies are 96.5%, 95.5% and 94.5%, respectively. Compared with other prediction models, the prediction accuracy is increased by 9.5%–18.5%, and this indicates the effectiveness of the prediction model DeepMal. Using deep learning network can enhance the robustness of DeepMal model for predicting malonylation sites. The source codes and all datasets are publicly available at https://github.com/QUST-AIBBDRC/DeepMal/. • A novel method (DeepMal) to predict the lysine malonylation sites of proteins. • Protein sequence information is extracted from sequence features, physicochemical properties, and evolutionary features. • Convolutional neural networks can learn to extract significant features associated with malonylation sites. • Deep neural networks can improve the accurate prediction of protein lysine malonylation sites of proteins. • The proposed method increases the prediction performance on independent test datasets compared with other methods. [ABSTRACT FROM AUTHOR]

As a reversible and widely existing post-translational modification of proteins, acetylation plays a crucial role in transcriptional regulation, apoptosis, and cytokine signaling. To better understand the molecular mechanism of acetylation, identification of acetylation sites is vital. The traditional experimental methods are time-consuming and cost-prohibitive, and the majority of acetylation sites remain unknown. It is necessary to develpoe an effective and accurate computational approach to predict the acetylation sites, especially utilizing the advanced deep learning technique. In this study, we propose a prokaryote acetylation sites prediction method based on the deep neural networks, named DNNAce. We extract the protein features from sequence information, physicochemical information, and evolution information of amino acid residues and obtain the initial feature set All. Moreover, we use Group Lasso to remove the irrelevant features that are not effective for classification in the sites prediction area. Compared to other machine learning methods, we use deep neural networks to predict acetylation sites. The 10-fold cross-validation on independent test datasets indicates that DNNAce has the highest values of accuracy for predicting acetylation sites. That DNNAce accurately identifies acetylation sites assists us in understanding its molecular mechanism, and provides related theoretical foundation for the development of drug targets for various diseases. The source code and all datasets are available at https://github.com/QUST-AIBBDRC/DNNAce/. • A novel method (DNNAce) to predict the prokaryote lysine acetylation sites in nine datasets. • The protein sequences are encoded by combining sequence information, physicochemical information and evolution information. • Group Lasso method remove the redundant and irrelevant feature to select the optimal feature subset for the first time. • Compared with other classifiers, DNNAce can extract abstract features from protein sequence through the hidden layers. • The proposed method increases the prediction performance on independent test datasets compared with other tools. [ABSTRACT FROM AUTHOR]