The contact map, a two-dimensional representation of three-dimensional protein structure, plays an important role in protein structure prediction because it brings crucial restraints on protein conformation exploration. In this chapter, we have summarized automatic methodology development for contact map prediction, whose models are generally categorized into three classes: correlated mutation analysis, direct-correlation analysis, and supervised learning models. The first two classes are unsupervised algorithms, and the last needs training samples extracted from experimentally solved protein structures. Protein residue contact prediction is an extremely imbalanced modeling problem in big data modeling, because the number of residue pairs increases exponentially with sequence length. It has hence triggered the recent deep learning model's successful applications in this topic. We also show in this chapter that the sequence-encoding features extracted from multiple sequence alignment are one of the keys for enhancing predictive performance. With the more accurate and faster models proposed for this challenging topic, predicted contact knowledge is expected to be capable of dramatically speeding up the protein 3-D structure prediction area by providing reliable and timely spatial restraints between residues.

Amphipathic helix (AH) features the segregation of polar and nonpolar residues and plays important roles in many membrane-associated biological processes through interacting with both the lipid and the soluble phases. Although the AH structure has been discovered for a long time, few ab initio machine learning-based prediction models have been reported, due to the limited amount of training data. In this study, we report a new deep learning-based prediction model, which is composed of a residual neural network and the uneven-thresholds decision algorithm. It is constructed on 121 membrane proteins, in total 51640 residue samples, which are curated from an up-to-date membrane protein structure database. Through a rigid 10-fold nested cross-validation experiment, we demonstrate that our model has exceeded the state-of-the-art approaches in this field. This presents a new avenue for accurately predicting AHs. Analysis on the contribution of the input residues and some cases further reveals the high interpretability and the generalization of our model.

A major challenge for protein databases is reconciling information from diverse sources. This is especially difficult when some information consists of secondary, human‐interpreted rather than primary data. For example, the Swiss‐Prot database contains curated annotations of subcellular location that are based on predictions from protein sequence, statements in scientific articles, and published experimental evidence. The Human Protein Atlas (HPA) consists of millions of high‐resolution microscopic images that show protein spatial distribution on a cellular and subcellular level. These images are manually annotated with protein subcellular locations by trained experts. The image annotations in HPA can capture the variation of subcellular location across different cell lines, tissues, or tissue states. Systematic investigation of the consistency between HPA and Swiss‐Prot assignments of subcellular location, which is important for understanding and utilizing protein location data from the two databases, has not been described previously. In this paper, we quantitatively evaluate the consistency of subcellular location annotations between HPA and Swiss‐Prot at multiple levels, as well as variation of protein locations across cell lines and tissues. Our results show that annotations of these two databases differ significantly in many cases, leading to proposed procedures for deriving and integrating the protein subcellular location data. We also find that proteins having highly variable locations are more likely to be biomarkers of diseases, providing support for incorporating analysis of subcellular location in protein biomarker identification and screening.

The rapid progress of cryo-electron microscopy (cryo-EM) in structural biology has raised an urgent need for robust methods to create and refine atomic-level structural models using low-resolution EM density maps. We propose a new protocol to create initial models using I-TASSER protein structure prediction, followed by EM density map-based rigid-body structure fitting, flexible fragment adjustment and atomic-level structure refinement simulations. The protocol was tested on a large set of 285 non-homologous proteins and generated structural models with correct folds for 260 proteins, where 28% had RMSDs below 2 Å. Compared to other state-of-the-art methods, the major advantage of the proposed pipeline lies in the uniform structure prediction and refinement protocol, as well as the extensive structural re-assembly simulations, which allow for low-to-medium resolution EM density map-guided structure modeling starting from amino acid sequences. Interestingly, the quality of both the image fitting and subsequent structure refinement was found to be strongly correlated with the correctness of the initial I-TASSER models; this is mainly due to the different correlation patterns observed between force field and structural quality for the models with template modeling score (or TM-score, a metric quantifying the similarity of models to the native) above and below a threshold of 0.5. Overall, the results demonstrate a new avenue that is ready to use for large-scale cryo-EM-based structure modeling and atomic-level density map-guided structure refinement.

Signal peptides play an important role in guiding and transferring transmembrane proteins and secreted proteins. In recent years, with the explosive growth of protein sequences, computationally predicting signal peptides and their cleavage sites from protein sequences is highly desired. In this work, we present an improved approach, Signal-3L 3.0, for signal peptide recognition and cleavage-site prediction using a 3-layer hybrid method of integrating deep learning algorithms and window-based scoring. There are three main components in the Signal-3L 3.0 prediction engine: (1) a deep bidirectional long short-term memory (Bi-LSTM) network with a soft self-attention learns abstract features from sequences to determine whether a query protein contains a signal peptide; (2) the statistics propensity window-based cleavage site screening method is applied to generate the set of candidate cleavage sites; (3) the prediction of a conditional random field with a hybrid convolutional neural network (CNN) and Bi-LSTM is fused with the window-based score for identifying the final unique cleavage site. Experimental results on the benchmark datasets show that the new deep learning-driven Signal-3L 3.0 yields promising performance. The online server of Signal-3L 3.0 is available at http://www.csbio.sjtu.edu.cn/bioinf/Signal-3L/.

RNA-binding protein (RBP) is a class of proteins that bind to and accompany RNAs in regulating biological processes. An RBP may have multiple target RNAs, and its aberrant expression can cause multiple diseases. Methods have been designed to predict whether a specific RBP can bind to an RNA and the position of the binding site using binary classification model. However, most of the existing methods do not take into account the binding similarity and correlation between different RBPs. While methods employing multiple labels and Long Short Term Memory Network (LSTM) are proposed to consider binding similarity between different RBPs, the accuracy remains low due to insufficient feature learning and multi-label learning on RNA sequences. In response to this challenge, the concept of RNA-RBP Binding Network (RRBN) is proposed in this paper to provide theoretical support for multi-label learning to identify RBPs that can bind to RNAs. It is experimentally shown that the RRBN information can significantly improve the prediction of unknown RNA−RBP interactions. To further improve the prediction accuracy, we present the novel computational method iDeepMV which integrates multi-view deep learning technology under the multi-label learning framework. iDeepMV first extracts data from the views of amino acid sequence and dipeptide component based on the RNA sequences as the original view. Deep neural network models are then designed for the respective views to perform deep feature learning. The extracted deep features are fed into multi-label classifiers which are trained with the RNA−RBP interaction information for the three views. Finally, a voting mechanism is designed to make comprehensive decision on the results of the multi-label classifiers. Our experimental results show that the prediction performance of iDeepMV, which combines multi-view deep feature learning models with RNA−RBP interaction information, is significantly better than that of the state-of-the-art methods. iDeepMV is freely available at http://www.csbio.sjtu.edu.cn/bioinf/iDeepMV for academic use. The code is freely available at http://github.com/uchihayht/iDeepMV.

In this study, we present an updated predictor DimiG 2.0, which uses a semi-supervised multi-label graph convolutional network (GCN) to infer disease-associated microRNAs (miRNAs) on an interaction network between protein coding genes (PCGs) and miRNAs using disease-PCG associations. DimiG 2.0 benefits from integrating the hierarchy of diseases into the GCN. DimiG 2.0 has the following updates: 1) It incorporates the hierarchy of diseases to regularize the GCN, encouraging diseases in the hierarchy to share similar miRNAs. 2) It integrates the PCGs with interacting partners but without associated diseases into model training, these unlabeled PCGs increase the size of the constructed interaction network. 3) It is able to predict associated miRNAs for 1017 diseases (updated from 248). 4) It updates expression data across tissues from the latest GTEx v7, and the expression values are quantified in Transcripts Per Million (TPM). Our results show that DimiG 2.0 outperforms state-of-the-art semi-supervised and supervised methods on the constructed benchmarked sets.

For the past decade, cryogenic electron microscopy (cryo-EM) has become an important technology to determine three-dimensional (3D) structures of biomacromolecules. Many software tools have been developed for cryo-EM image processing and 3D reconstruction, covering various computational tasks in cryo-EM data analysis. Despite the recent progress, most of these software tools focus on a single task, such as automatic particle picking or image clustering, whereas software packages covering the whole pipeline of cryo-EM data processing are still few. In this study, we developed a fully automatic single-particle reconstruction and analysis toolkit for cryo-EM data, named SPREAD, which integrates 2D image classification, 3D initial model generation, model selection, and 3D refinement. In SPREAD, we adopt our previously proposed network-based clustering algorithm for 2D image classification, NCEM, and the reference-free resolution measurement method SRes to realize the automatic model ranking and selection procedure. Projection orientation assignment is one of the key steps in initial model generation and 3D refinement. In SPREAD, we use the network-based image similarity metric and introduce a new probabilistic-based orientation searching method, named peak finding, to enhance assignment of the projection orientations. For dealing with both the particle images and projection images in the 3D refinement using SPREAD, we build a mixture image network containing both of these types of images on the basis of the peak-finding results, and then similarities for node pairs are recomputed by a superposed random walk on the network. SPREAD achieves a fully automatic workflow in which nearly no expert domain knowledge and interactive manual operation are involved. Our software can accessed for free at http://www.csbio.sjtu.edu.cn/bioinf/SPREAD/ for academic use.

Motivation Knowledge of protein–ligand binding residues is important for understanding the functions of proteins and their interaction mechanisms. From experimentally solved protein structures, how to accurately identify its potential binding sites of a specific ligand on the protein is still a challenging problem. Compared with structure-alignment-based methods, machine learning algorithms provide an alternative flexible solution which is less dependent on annotated homogeneous protein structures. Several factors are important for an efficient protein–ligand prediction model, e.g. discriminative feature representation and effective learning architecture to deal with both the large-scale and severely imbalanced data. Results In this study, we propose a novel deep-learning-based method called DELIA for protein–ligand binding residue prediction. In DELIA, a hybrid deep neural network is designed to integrate 1D sequence-based features with 2D structure-based amino acid distance matrices. To overcome the problem of severe data imbalance between the binding and nonbinding residues, strategies of oversampling in mini-batch, random undersampling and stacking ensemble are designed to enhance the model. Experimental results on five benchmark datasets demonstrate the effectiveness of proposed DELIA pipeline.

Overnutrition results in adiposity and chronic inflammation with expansion of white adipose tissue (WAT). However, genetic factors controlling fat mass and adiposity remain largely undetermined. We applied whole-exome sequencing in young obese subjects and identified rare gain-of-function mutations in CTNNB1/β-catenin associated with increased obesity risk. Specific ablation of β-catenin in mature adipocytes attenuated high-fat diet–induced obesity and reduced sWAT mass expansion with less proliferated Pdgfrα+ preadipocytes and less mature adipocytes. Mechanistically, β-catenin regulated the transcription of serum amyloid A3 (Saa3), an adipocyte-derived chemokine, through β-catenin–TCF (T-Cell-Specific Transcription Factor) complex in mature adipocytes, and Saa3 activated macrophages to secrete several factors, including Pdgf-aa, which further promoted the proliferation of preadipocytes, suggesting that β-catenin/Saa3/macrophages may mediate mature adipocyte-preadipocyte cross-talk and fat expansion in sWAT. The identification of β-catenin as a key regulator in fat expansion and human adiposity provides the basis for developing drugs targeting Wnt/β-catenin pathway to combat obesity.