Protein structure prediction is important for understanding their function and behavior. This review study presents a comprehensive review of the computational models used in predicting protein structure. It covers the progression from established protein modeling to state-of-the-art artificial intelligence (AI) frameworks. The paper will start with a brief introduction to protein structures, protein modeling, and AI. The section on established protein modeling will discuss homology modeling, ab initio modeling, and threading. The next section is deep learning-based models. It introduces some state-of-the-art AI models, such as AlphaFold (AlphaFold, AlphaFold2, AlphaFold3), RoseTTAFold, ProteinBERT, etc. This section also discusses how AI techniques have been integrated into established frameworks like Swiss-Model, Rosetta, and I-TASSER. The model performance is compared using the rankings of CASP14 (Critical Assessment of Structure Prediction) and CASP15. CASP16 is ongoing, and its results are not included in this review. Continuous Automated Model EvaluatiOn (CAMEO) complements the biennial CASP experiment. Template modeling score (TM-score), global distance test total score (GDT_TS), and Local Distance Difference Test (lDDT) score are discussed too. This paper then acknowledges the ongoing difficulties in predicting protein structure and emphasizes the necessity of additional searches like dynamic protein behavior, conformational changes, and protein–protein interactions. In the application section, this paper introduces some applications in various fields like drug design, industry, education, and novel protein development. In summary, this paper provides a comprehensive overview of the latest advancements in established protein modeling and deep learning-based models for protein structure predictions. It emphasizes the significant advancements achieved by AI and identifies potential areas for further investigation.

The precise identification of drug-protein inter action (DPI) can significantly speed up the drug discovery process. Bioassay methods are time-consuming and expensive to screen for each pair of drug proteins. Machine-learning-based methods cannot accurately predict a large number of DPIs. Compared with traditional computing methods, deep learning methods need less domain knowledge and have strong data learning ability. In this study, we construct a DPI prediction model based on dual channel neural networks with an efficient path attention mechanism, called DCA-DPI. The drug molecular graph and protein sequence are used as the data input of the model, and the residual graph neural network and the residual convolution network are used to learn the feature representation of the drug and protein, respectively, to obtain the feature vector of the drug and the hidden vector of protein. To get a more accurate protein feature vector, the weighted sum of the hidden vector of protein is applied using the neural attention mechanism. In the end, drug and protein vectors are concatenated and input into the full connection layer for classification. In order to evaluate the performance of DCA-DPI, three widely used public data, Human, C.elegans and DUD-E, are used in the experiment. The evaluation metrics values in the experiment are superior to other relevant methods. Experiments show that our model is efficient for DPI prediction.

In late 2019, the emergence of a novel coronavirus led to its identification as SARS-CoV-2, precipitating the onset of the COVID-19 pandemic. Many experimental and computational studies were performed on SARS-CoV-2 to understand its behavior and patterns. In this research, Molecular Dynamic (MD) simulation is utilized to compare the behaviors of SARS-CoV-2 and its Variants of Concern (VOC)-Alpha, Beta, Gamma, Delta, and Omicron-with the hACE2 protein. Protein structures from the Protein Data Bank (PDB) were aligned and trimmed for consistency using Chimera, focusing on the receptor-binding domain (RBD) responsible for ACE2 interaction. MD simulations were performed using Visual Molecular Dynamics (VMD) and Nanoscale Molecular Dynamics (NAMD2), and salt bridges and hydrogen bond data were extracted from the results of these simulations. The data extracted from the last 5 ns of the 10 ns simulations were visualized, providing insights into the comparative stability of each variant’s interaction with ACE2. Moreover, electrostatics and hydrophobic protein surfaces were calculated, visualized, and analyzed. Our comprehensive computational results are helpful for drug discovery and future vaccine designs as they provide information regarding the vital amino acids in protein-protein interactions (PPIs). Our analysis reveals that the Original and Omicron variants are the two most structurally similar proteins. The Gamma variant forms the strongest interaction with hACE2 through hydrogen bonds, while Alpha and Delta form the most stable salt bridges; the Omicron is dominated by positive potential in the binding site, which makes it easy to attract the hACE2 receptor; meanwhile, the Original, Beta, Delta, and Omicron variants show varying levels of interaction stability through both hydrogen bonds and salt bridges, indicating that targeted therapeutic agents can disrupt these critical interactions to prevent SARS-CoV-2 infection.

This paper first studies the generalization ability of the convolutional layer as a feature mapper (CFM) for extracting image features and the classification ability of the multilayer perception (MLP) in a CNN. Then, a novel generalized hybrid probability convolutional neural network (GHP-CNN) is proposed to solve abstract feature classification with an unknown distribution form. To measure the generalization ability of the CFM, a new index is defined and the positive correlation between it and the CFM is researched. Generally, a fully trained CFM can extract features that are beneficial to classification, regardless of whether the data participate in training the CFM. In the CNN, the fully connected layer in the MLP is not always optimal, and the extracted abstract feature has an unknown distribution. Thus, an improved classifier called the structure-optimized probabilistic neural network (SOPNN) is used for abstract feature classification in the GHP-CNN. In the SOPNN, the separability information is not lost in the normalization process, and the final classification surface is close to the optimal classification surface under the Bayesian criterion. The proposed GHP-CNN utilizes the generalization ability of the CFM and the classification ability of the SOPNN. Experiments show that the proposed network has better classification ability than the existing hybrid neural networks.

Medical image analysis and classification, using machine learning, particularly Convolutional Neural Networks, have demonstrated a great deal of success. Research into mammography image classification tended to focus on either binary outcome (malignancy or benign) or nominal (unordered) classification for multiclass labels [1]. The industry standard metric for radiologist's classification of mammography images is a rating scale called BI-RADS (Breast Imaging Reporting and Data System), where values 1 through 5 are a distinct progression of assessment that are intended to denote higher risk of a malignancy, based on the characteristics of anomalies within an image [1][2][3]. The development of a classifier that predicts BI-RADS 1-5, would provide radiologists with an objective second opinion on image anomalies. In this paper, we applied a novel Deep Learning method called OHPLall (Ordinal Hyperplane Loss - all centroids), which was specifically designed for data with ordinal classes, to the predictions of BI-RADS scales on mammography images. Our experimental study demonstrated promising results generated by OHPLall and great potential of using OHPLall models as a supplemental diagnostic tool.

Net Promotor Score is an important business measurement process where customers are surveyed and asked to rate their likelihood of recommending the company's products and/or services. In many applications, customers are asked to respond on an 11-point ordinal scale of 0 to 10. In developing the score, the data are reformulated into a labelled 3 class scale (0-6: Detractor, 7-8: Passive and 9-10: Promoter). [1] Many companies that choose to use Net Promoter Score as a core management metric integrate the measurement into all phases of the company and seek every opportunity to assess company performance in terms of likelihood to promote the company. In addition to a variety of survey opportunities, the ability to score comments in survey, social media and blogs with promoter rating may provide an additional valuable source of business insight. Even on a three-point scale, Net Promoter is an ordinal classification problem. A number of successful algorithms, that develop ordinal classifiers have been developed. [2] None of the top performing classifiers can be used for applications like text classification or image classification, since they don't employ deep learning. Any appropriate strategy must utilize the ordering information of classes without imposing a strong continuous assumption or fixed spacing assumption on the ordinal classes. In this paper, we use a novel Deep Learning methodology called OHPLnet (Ordinal Hyperplane Loss Network) that is specifically designed for data with ordinal classes. [3] The algorithm is used to develop predictions of the eleven classes, that may be used in the standard Net Promoter Score generation process.

Deep learning has evolved into one of the most powerful techniques for analytics on both structured and unstructured data. As a well-adopted analytics system, SAS has also integrated deep learning functionalities into its product family, such as SAS Viya, SAS Cloud Analytic Services, and SAS Visual Data Mining and Machine Learning. In this paper, we conduct an in-depth comparison between SAS and Python on their deep learning modeling with different types of data, including structured, images, text, and sequential data. We focus on using such deep learning frameworks in SAS environment, and highlight the main differences between SAS and Python on programming styles on deep learning along with each tool’s advantages and disadvantages.

The electronic health record (EHR) analysis has become an increasingly important landing area for machine learning and text mining algorithms to leverage the full potential of the big data for improving human health care. In a lot of our Chinese EHR analysis applications, it is very important to categorize the patients' diseases according to the Chinese national medical coding standard. In this paper, we develop NLP and machine learning algorithms to automatically categorize each patient's diseases into one or more categories. We take each patient's disease description as a document. Also, for each disease category, we make use of its description information in the medical coding standard and take it as a document. According to the characteristics of our data, we define the categorization problem as the unsupervised classification problem with the nearest neighborhood (NN) algorithm using different vector representations to represent the documents. Experimental results show that the averaged word embeddings of word2Vec works best with very promising classification performance.

The problem of ordinal classification occurs in a large and growing number of areas. Some of the most common source and applications of ordinal data include rating scales, medical classification scales, socio-economic scales, meaningful groupings of continuous data, facial emotional intensity, facial age estimation, etc. The problem of predicting ordinal classes is typically addressed by either performing n-1 binary classification for n ordinal classes or treating ordinal classes as continuous values for regression. However, the first strategy doesn't fully utilize the ordering information of classes and the second strategy imposes a strong continuous assumption to ordinal classes. In this paper, we propose a novel loss function called Ordinal Hyperplane Loss (OHPL) that is particularly designed for data with ordinal classes. The proposal of OHPL is a significant advancement in predicting ordinal class data, since it enables deep learning techniques to be applied to the ordinal classification problem on both structured and unstructured data. By minimizing OHPL, a deep neural network learns to map data to an optimal space where the distance between points and their class centroids are minimized while a nontrivial ordinal relationship among classes are maintained. Experimental results show that deep neural network with OHPL not only outperforms the state-of-the-art alternatives on classification accuracy but also scales well to large ordinal classification problems.

The rapid development of social networking sites brings about many data mining tasks and novel challenges. We focus on classification tasks with students’ interaction information in a social network. To mitigate the difficulties of developing a learning system, this study proposes a new computing paradigm: spectral clustering as a service, providing a service to enable exacting social dimensionality on demand. Spectral clustering has been developed in a social network dimensionality refinement model as a kernel middleware, namely SNDR. The SNDR service can process the sparse information, explore the network’s topology and finally exact suitable features. Experimental results justify the design of Collective Behavior Learning System and the implementation of the Social Network Dimensionality Refinement model’s service. Our system makes better performance than baseline methods.

Summary Dynamic time warping (DTW) has been widely used in various pattern recognition and time series data mining applications. However, as examples will illustrate, both the classic DTW and its later alternative, derivative DTW, may fail to align a pair of sequences on their common trends or patterns. Furthermore, the learning capability of any supervised learning algorithm based on classic/derivative DTW is very limited. In order to capture trends or patterns that a sequence presents during the alignment process, we first derive a global feature and a local feature for each point in a sequence. Then, a method called feature based dynamic time warping (FBDTW) is designed to align two sequences based on each point‟s local and global features instead of its value or derivative. Experimental study shows that FDBTW outperforms both classic DTW and derivative DTW on pairwise distance evaluation of time series sequences. In order to enhance the capacity of supervised learning based on DTW, we further design a method called adaptive feature based dynamic time warping (AFDBTW) by equipping the FDBTW with a novel feature selection algorithm. This feature selection algorithm is able to expand the learning capability of any DTW based supervised learning algorithm by a dual learning process. The first-fold learning process learns the significances of both the local feature and global feature towards classification; then the second-fold learning process learns a classification model based on the pairwise distances generated by the AFDBTW. A comprehensive experimental study shows that the AFDBTW is able to make further improvement over the FDBTW in time series classification.