Representation Of Sequences Research Articles

Case scenario generators for trauma surgery simulation utilizing autoregressive language models

Trauma is the leading cause of death in adults under the age of 45 and the fourth leading cause of death in the United States. Effective delivery of trauma care centers on being well versed in the Advanced Trauma Life Support (ATLS) protocol, which requires high levels of clinical experience. Often this comes from having been exposed to the many permutations of common types of injuries as well as exposed to rarer scenarios, but with potential harm to patients. Case scenarios, which are sequential representations of clinical events, can help trainees receive clinical exposure without harming patients. However authoring case scenarios requires domain expertise, wide experience, and the ability to intelligently respond to inputs, and as such is currently an arduous task. Autoregressive generative models trained on large amounts of clinical data, such as the National Trauma Data Bank (NTDB), pose a possible solution to overcome the cost of authorship while providing broad and accessible clinical experience to trainees. We have developed a Trauma AI model composed of an autoregressive generative model based on the transformer architecture for generating potential case scenario combined with an out-of-domain detection for filtering out less plausible scenarios. The GPT2 model is trained on 1.1 million case scenarios derived from the NTDB data. We demonstrate that Trauma AI is capable of generating realistic case scenarios that encode the ATLS protocol as a latent feature of the sequence of provider interventions, including scenarios that do not have any parallels in the original dataset. We also present an unsupervised means of filtering out unrealistic sequences by identifying out-of-domain sequences, and demonstrate that this improves the realism of the generated case scenarios.

HGV4Risk: Hierarchical Global View-guided Sequence Representation Learning for Risk Prediction

Risk prediction, usually achieved by learning representations from patient’s physiological sequence or user’s behavioral sequence data, and has been widely applied in healthcare and finance. Despite that, some recent time-aware deep learning methods have led to superior performances in such sequence representation learning tasks, such improvement is limited due to a lack of guidance from hierarchical global view. To address this issue, we propose a novel end-to-end H ierarchical G lobal V iew-guided (HGV) sequence representation learning framework. Specifically, the Global Graph Embedding (GGE) module is proposed to learn sequential clip-aware representations from temporal correlation graph (TCG) at instance level. Furthermore, following the way of key-query attention, the harmonic β-attention (β-Attn) is also developed for making a global tradeoff between time-aware decay and observation significance at channel level adaptively. Moreover, the hierarchical representations at both instance level and channel level can be coordinated by the heterogeneous information aggregation under the guidance of global view. Experimental results on both healthcare risk prediction benchmark and SMEs credit overdue risk prediction task from the real-world industrial scenario in MYBank, Ant Group, have illustrated that the proposed model can achieve competitive prediction performance compared with other known baselines. The code has been released public available at: https://github.com/LiYouru0228/HGV.

Dynamic graph neural network-based fraud detectors against collaborative fraudsters

Telecom fraud detection is a challenging task since the fact that fraudulent behaviors are hidden in the vast amount of telecom records. More concerning, the ongoing coronavirus pandemic (COVID-19) accelerated the use of mobile internet, providing more criminal opportunities for fraudsters. However, current telecom fraud detection mostly focuses on individual sequences representation, rarely noticing the collaboration of fraudsters, making it exhibit unsatisfactory performance in the face of gang crimes. To address this problem, we propose to extract collaborative networks from user call logs with an emphasis on unveiling collaborative fraud. We employ eight months of telecom datasets in China with 6,106 users and 5.0 million call logs between 1.25 million telephone recipients. Through our study, we find that the social structure of fraudsters evolute rapidly while the normal users remain stable relatively. In addition, we find that mining collaborative fraud strategies help to detect fraudsters with less distinct fraud characteristics. To this end, we propose a novel model named COllaborative-REsistant Dynamic Graph Neural Network (CORE-DGNN), to enhance the dynamic GNN aggregation process. Specifically, we first use co-recipients to obtain the collaborative network under each time slice. Then, we design a multi-frequency graph neural network to adaptively aggregate the features of node neighbors at different frequencies to address the problem of heterophily in collaborative networks. Finally, a self-attentive temporal convolutional network is designed to aggregate node embedding features across multiple time spans. Comprehensive experiments on two real-world telecom fraud datasets show that our approach outperforms several state-of-the-art algorithms.

Structural Vibration Data Anomaly Detection Based on Multiple Feature Information Using CNN-LSTM Model

Structural health monitoring (SHM) system has been operating for a long time in a harsh environment, resulting in various abnormalities in the collected structural vibration monitoring data. Detecting these abnormal data not only requires user interaction but also is quite time-consuming. Inspired by the manual recognition process, a vibration data anomaly detection method based on the combined model of convolutional neural network (CNN) and long short-term memory (LSTM) network is proposed in this paper. This method simulates intelligent human decision making in two steps. First, the original data are reconstructed by two feature sequences with higher universality and smaller size. In the time domain, the residual signal is extracted from the upper and lower peak envelopes of the original data to characterize the symmetry of the data. In the frequency domain, the power spectral density sequence of the original data is extracted to characterize the interpretability of the data. Second, a CNN-LSTM model is constructed and trained which utilizes CNN to extract local high-level features of input sequence and inputs new continuous high-level feature representations into LSTM to learn global long-term dependencies of abnormal data features. For verification, the method was applied to the automatic classification of continuous monitoring data for 42 days of long-span bridge, and the average accuracy of the classification results exceeded 94% and the detection time was 78 minutes. Compared with existing methods, this method can detect abnormal data more accurately and efficiently and has a stronger generalization ability.

CMMS-GCL: Cross-Modality Metabolic Stability Prediction with Graph Contrastive Learning.

Metabolic stability plays a crucial role in the early stages of drug discovery and development. Accurately modeling and predicting molecular metabolic stability has great potential for the efficient screening of drug candidates as well as the optimization of lead compounds. Considering wet-lab experiment is time-consuming, laborious, and expensive, in silico prediction of metabolic stability is an alternative choice. However, few computational methods have been developed to address this task. In addition, it remains a significant challenge to explain key functional groups determining metabolic stability. To address these issues, we develop a novel cross-modality graph contrastive learning model named CMMS-GCL for predicting the metabolic stability of drug candidates. In our framework, we design deep learning methods to extract features for molecules from two modality data, i.e., SMILES sequence and molecule graph. In particular, for the sequence data, we design a multi-head attention BiGRU-based encoder to preserve the context of symbols to learn sequence representations of molecules. For the graph data, we propose a graph contrastive learning-based encoder to learn structure representations by effectively capturing the consistencies between local and global structures. We further exploit fully connected neural networks to combine the sequence and structure representations for model training. Extensive experimental results on two datasets demonstrate that our CMMS-GCL consistently outperforms seven state-of-the-art methods. Furthermore, a collection of case studies on sequence data and statistical analyses of the graph structure module, strengthens the validation of the interpretability of crucial functional groups recognized by CMMS-GCL. Overall, CMMS-GCL can serve as an effective and interpretable tool for predicting metabolic stability, identifying critical functional groups, and thus facilitating the drug discovery process and lead compound optimization. The code and data underlying this article are freely available at https://github.com/dubingxue/CMMS-GCL. Supplementary data are available at Bioinformatics online.

Impaired extraction and consolidation of morphological regularities in developmental dyslexia: A domain general deficit?

The current study examined whether adults with Developmental Dyslexia are impaired in learning linguistic regularities in a novel language, and whether this may be explained by a domain general deficit in the effect of sleep on consolidation. We compared online learning and offline consolidation of morphological regularities in individuals with Developmental Dyslexia (N = 40) and typical readers (N = 38). Participants learned to apply plural inflections to novel words based on morpho-phonological rules embedded in the input and learned to execute a finger motor sequence task. To test the effects of time and sleep on consolidation, participants were assigned into one of two sleep-schedule groups, trained in the evening or in the morning and tested 12 and 24 h later. Unlike typical readers, Dyslexic readers did not extract the morpho-phonological regularities during training and as a group they did not show offline gains in inflecting trained items 24 h after training, suggesting that the deficit in extraction of regularities during training may be related to the deficit in consolidation. The offline gains in dyslexic readers, were correlated with their prior phonological abilities, and were less affected by sleep than those of typical readers. Although no deficit was found in the consolidation of the motor task, dyslexic readers were again less successful in generating an abstract representation of the motor sequence, reflected in a difficulty to generalize the motor sequence knowledge acquired using one hand to the untrained hand. The results suggest that individuals with Developmental Dyslexia have a domain general deficit in extracting statistical regularities from an input. Within the language domain this deficit is reflected in reduced benefits of consolidation, particularly during sleep, perhaps due to reduced prior phonological abilities, which may impede the individual's ability to extract the linguistic regularities during and after training and thus constrain the consolidation process.

Higher order Morita approximation and its validity for random copolymer adsorption onto homogeneous and periodic heterogeneous surfaces

Adsorption of a single AB random copolymer (RC) chain onto homogeneous and inhomogeneous ab surfaces with a regular periodic pattern is studied theoretically. For the averaging over disorder in the RC sequence, the constrained annealed approximation, known as the Morita approximation is employed. A general scheme for constructing the Morita approximation of an arbitrary order m is proposed; it is based on representation of the RC monomer sequence as a Markov chain of overlapping –1)-ads. The problem is solved within the framework of the generating functions approach for the two-dimensional partially directed walk model of the polymer on a square lattice. Temperature dependences of various adsorption characteristics are obtained. The accuracy of the Morita approximation is assessed by comparison with the numerical results for RC with quenched sequences obtained by the transfer matrix approach. It is shown that for RC adsorption on a homogeneous surface for AB-copolymers with adsorbing A and neutral B blocks, it is sufficient to use the Morita approximation of second or third order. If the non-adsorbing B block is repelled from the surface, then a higher order of the approximation (5th–6th) is required. An important indicator of validity of the Morita approximation is the entropy: if the order of the approximation is not high enough, the entropy becomes negative in the low-temperature regime which means that the Morita approximation is wrong in this order. In the case when the surface is periodically inhomogeneous, the Morita approximation works worse and very high orders are required, which can be beyond the computational capabilities. Moreover, the Morita approximations become degenerate: approximations of two or more consecutive orders give the same result.

Effects of heated tobacco product aerosol extracts on DNA methylation and gene transcription in lung epithelial cells

AimsSmoking causes DNA methylation (DNAm) alterations that lead to lung cancer development. Although the use of heated tobacco products (HTPs) has recently increased, their impact on health remains unclear. This study aimed to evaluate the effects of HTPs on DNAm and gene transcription in human lung epithelial cells in vitro. Main methodsHuman lung adenocarcinoma (A549) cells with type II alveolar epithelial characteristics were treated with aerosol extracts of two HTPs or a smoke extract of combustible reference cigarette (RC). Global 5-methylcytosine (5-mC) and 5-hydroxymethylcytosine (5-hmC) levels were quantified using dot blot analysis. Furthermore, reduced representation of bisulfite sequencing, DNA microarray, and quantitative PCR analyses were performed to determine CpG methylation and gene transcription changes induced by HTP and RC. Key findingsGlobal 5-mC and 5-hmC levels were decreased by the RC extract but not the HTP extracts. However, an HTP extract altered the CpG methylation pattern, and Gene Ontology enrichment analysis of the differentially methylated regions of the RC and HTP groups showed a similar pattern. The HTP extract affected gene expression, albeit to a lesser extent than the RC extract. In particular, the HTP extract markedly affected the mRNA expression and promoter methylation of cytochrome P450 family 1 subfamily A member 1 (CYP1A1), which is associated with carcinogenic risk. SignificanceThe study results suggest that HTPs as well as conventional combustible cigarettes can alter CpG methylation and gene transcription in lung epithelial cells.

MATC-Net: Learning compact sequence representation for hierarchical loop closure detection

Loop closure detection (LCD) is a challenging task to judge whether the current position of an intelligent robot returns to the previously visited position. Mainstream appearance-based approaches apply robust image representation techniques to describe the scene. However, most of these methods are designed for single images, and the sequence representation method incorporating temporal sequence information is still in the preliminary exploration. In this paper, we propose a compact sequence representation method for hierarchical LCD, ensuring conspicuous performance for the LCD task. Deriving from a group of global features of image sequences, we propose a multi-scale asymmetric temporal convolution network (MATC-Net), which generates sequential features and transformed global features through its aggregation branch and transformation branch, respectively. Based on these two types of features, a MATC-Net-based hierarchical LCD framework including two similarity measurement processes is constructed, through which the best place match is identified. The experimental results show that our method outperforms other counterparts on three datasets, exhibiting the promising potential of leveraging sequential features to LCD task.

Detection of Facial Mask Using Deep Learning Classification Algorithm

Deep learning is an algorithm that works by representing data in layers of learning layers so that the representation becomes more meaningful. "Deep" in deep learning means that deep learning begins layers of sequential representation. This study aims to provide a reference on how to create a system and analyze the results of identifying face masks using a deep learning algorithm. Research on face mask detection is highly important as it tackles a vital element of public health and safety. It plays a crucial role in promoting adherence to mask-wearing guidelines, minimizing the transmission of infectious diseases, and offering valuable data for monitoring and policy assessment. Additionally, this area of study has garnered increased significance and attention within the realm of public health and safety, especially in light of the COVID-19 pandemic since 2020, where mask usage has been universally advised to protect individuals from the spread of the virus. From the results of research conducted, it is known that this model can recognize faces well, both those who wear masks and those who do not use masks. This is evident from the average specificity and precision of 96.00%, and the average sensitivity or recall value of 93.47%. In addition, this model has also proven to be quite accurate in conducting overall classification with an average accuracy of 94.73%.

Zero-shot-capable identification of phage-host relationships with whole-genome sequence representation by contrastive learning.

Accurately identifying phage-host relationships from their genome sequences is still challenging, especially for those phages and hosts with less homologous sequences. In this work, focusing on identifying the phage-host relationships at the species and genus level, we propose a contrastive learning based approach to learn whole-genome sequence embeddings that can take account of phage-host interactions (PHIs). Contrastive learning is used to make phages infecting the same hosts close to each other in the new representation space. Specifically, we rephrase whole-genome sequences with frequency chaos game representation (FCGR) and learn latent embeddings that 'encapsulate' phages and host relationships through contrastive learning. The contrastive learning method works well on the imbalanced dataset. Based on the learned embeddings, a proposed pipeline named CL4PHI can predict known hosts and unseen hosts in training. We compare our method with two recently proposed state-of-the-art learning-based methods on their benchmark datasets. The experiment results demonstrate that the proposed method using contrastive learning improves the prediction accuracy on known hosts and demonstrates a zero-shot prediction capability on unseen hosts. In terms of potential applications, the rapid pace of genome sequencing across different species has resulted in a vast amount of whole-genome sequencing data that require efficient computational methods for identifying phage-host interactions. The proposed approach is expected to address this need by efficiently processing whole-genome sequences of phages and prokaryotic hosts and capturing features related to phage-host relationships for genome sequence representation. This approach can be used to accelerate the discovery of phage-host interactions and aid in the development of phage-based therapies for infectious diseases.

Knowledge-enhanced event relation extraction via event ontology prompt

Identifying temporal and subevent relationships between different events (i.e., event relation extraction) is an important step towards event-centric natural language processing, which can help understand how events evolve and potentially facilitate many downstream tasks, such as timeline generation and event knowledge graph construction. Existing work has extensively leveraged external knowledge to improve the performance of relation extraction. Despite the progress made, the current knowledge-enhanced approach still has some shortcomings, e.g., knowledge missing, knowledge noise, and suboptimal knowledge injection. In this paper, we propose OntoEnhance, a novel event relation extraction framework that fuses semantic information from event ontologies to enhance event representation. Specifically, we first inject the latent knowledge in the event ontology into the prompt text to address the issue of knowledge missing. Then a dual-stack attention fusion mechanism is further introduced to enhance the injection of key knowledge to alleviate knowledge noise. In order to prevent the knowledge in the event ontology from being wrongly dominated, we use the event direction induction mechanism to obtain the event context-based relational sequence representation. Finally, a gate mechanism is used to fuse ontology-based knowledge and context-based event features dynamically. Extensive experiments demonstrate that OntoEnhance outperforms all comparison baselines by a large margin on all four datasets under both standard and few-shot settings.

Rm-LR: A long-range-based deep learning model for predicting multiple types of RNA modifications

Recent research has highlighted the pivotal role of RNA post-transcriptional modifications in the regulation of RNA expression and function. Accurate identification of RNA modification sites is important for understanding RNA function. In this study, we propose a novel RNA modification prediction method, namely Rm-LR, which leverages a long-range-based deep learning approach to accurately predict multiple types of RNA modifications using RNA sequences only. Rm-LR incorporates two large-scale RNA language pre-trained models to capture discriminative sequential information and learn local important features, which are subsequently integrated through a bilinear attention network. Rm-LR supports a total of ten RNA modification types (m6A, m1A, m5C, m5U, m6Am, Ψ, Am, Cm, Gm, and Um) and significantly outperforms the state-of-the-art methods in terms of predictive capability on benchmark datasets. Experimental results show the effectiveness and superiority of Rm-LR in prediction of various RNA modifications, demonstrating the strong adaptability and robustness of our proposed model. We demonstrate that RNA language pretrained models enable to learn dense biological sequential representations from large-scale long-range RNA corpus, and meanwhile enhance the interpretability of the models. This work contributes to the development of accurate and reliable computational models for RNA modification prediction, providing insights into the complex landscape of RNA modifications.

Linear complexity and trace representation of balanced quaternary cyclotomic sequences of prime period p

Linear complexity and trace representation of balanced quaternary cyclotomic sequences of prime period p

SpecTrHuMS: Spectral transformer for human mesh sequence learning

We present SpecTrHuMS, a Spectral Transformer for 3D triangular Human Mesh Sequence learning which combines known deep learning models with spectral mesh processing to capture characteristics of 3D shapes as well as temporal dependencies between the frames. Unlike previous works in this field, our approach is able to work directly with a compressed representation of the geometry, the spectral coefficients, rather than relying solely on skeleton joints that does not contain surface information. The vertices of each mesh of a sequence are first projected on the eigenvectors of the Graph Laplacian computed from the common triangulation. A convolutional encoder then encodes each frame into lower dimensional latent variables that preserve as much as possible the spectral information. These latent variables are next passed through a transformer architecture so that the model understands the context of the sequence and learns temporal dependencies between the frames. Each frame of the transformer’s output is then decoded by a convolutional decoder which aims to reconstruct the input spectral coefficients. Finally, all frames are transformed back into the spatial domain, resulting in a general process able to treat 4D surfaces with a constant connectivity. Our method is evaluated on a prediction task on AMASS, a dataset of human surface sequences, showing the ability of our model to produce realistic movements while preserving the identity of a subject, and showing that this work is a significant step towards efficient and high-quality representation of triangular mesh sequences with constant connectivity. Additional experiments show that our model can be easily extended to other tasks such as long term prediction, completion and that it is generalizable to other datasets with constant connectivity. This work opens up new possibilities for applications in the fields of animation, virtual reality, and computer graphics. Pretrained models, the code to train them and the code to create datasets will be made publicly available.

Assessment of Genetic Variability and Evolutionary Relationships of Rhizoctonia solani Inherent in Legume Crops.

Rhizoctonia solani is one of the most common soil-borne fungal pathogens of legume crops worldwide. We collected rDNA-ITS sequences from NCBI GenBank, and the aim of this study was to examine the genetic diversity and phylogenetic relationships of various R. solani anastomosis groups (AGs) that are commonly associated with grain legumes (such as soybean, common bean, pea, peanut, cowpea, and chickpea) and forage legumes (including alfalfa and clover). Soybean is recognized as a host for multiple AGs, with AG-1 and AG-2 being extensively investigated. This is evidenced by the higher representation of sequences associated with these AGs in the NCBI GenBank. Other AGs documented in soybean include AG-4, AG-7, AG-11, AG-5, AG-6, and AG-9. Moreover, AG-4 has been extensively studied concerning its occurrence in chickpea, pea, peanut, and alfalfa. Research on the common bean has been primarily focused on AG-2, AG-4, and AG-1. Similarly, AG-1 has been the subject of extensive investigation in clover and cowpea. Collectively, AG-1, AG-2, and AG-4 have consistently been identified and studied across these diverse legume crops. The phylogenetic analysis of R. solani isolates across different legumes indicates that the distinct clades or subclades formed by the isolates correspond to their specific anastomosis groups (AGs) and subgroups, rather than being determined by their host legume crop. Additionally, there is a high degree of sequence similarity among isolates within the same clade or subclade. Principal coordinate analysis (PCoA) further supports this finding, as isolates belonging to the same AGs and/or subgroups cluster together, irrespective of their host legume. Therefore, the observed clustering of R. solani AGs and subgroups without a direct association with the host legume crop provides additional support for the concept of AGs in understanding the genetic relationships and evolution of R. solani.

Prediction of Antigenic Distance in Influenza A Using Attribute Network Embedding.

Owing to the rapid changes in the antigenicity of influenza viruses, it is difficult for humans to obtain lasting immunity through antiviral therapy. Hence, tracking the dynamic changes in the antigenicity of influenza viruses can provide a basis for vaccines and drug treatments to cope with the spread of influenza viruses. In this paper, we developed a novel quantitative prediction method to predict the antigenic distance between virus strains using attribute network embedding techniques. An antigenic network is built to model and combine the genetic and antigenic characteristics of the influenza A virus H3N2, using the continuous distributed representation of the virus strain protein sequence (ProtVec) as a node attribute and the antigenic distance between virus strains as an edge weight. The results show a strong positive correlation between supplementing genetic features and antigenic distance prediction accuracy. Further analysis indicates that our prediction model can comprehensively and accurately track the differences in antigenic distances between vaccines and influenza virus strains, and it outperforms existing methods in predicting antigenic distances between strains.

Contrastive Pre-training with Adversarial Perturbations for Check-In Sequence Representation Learning

A core step of mining human mobility data is to learn accurate representations for user-generated check-in sequences. The learned representations should be able to fully describe the spatial-temporal mobility patterns of users and the high-level semantics of traveling. However, existing check-in sequence representation learning is usually implicitly achieved by end-to-end models designed for specific downstream tasks, resulting in unsatisfactory generalizable abilities and poor performance. Besides, although the sequence representation learning models that follow the contrastive learning pre-training paradigm have achieved breakthroughs in many fields like NLP, they fail to simultaneously consider the unique spatial-temporal characteristics of check-in sequences and need manual adjustments on the data augmentation strategies. So, directly applying them to check-in sequences cannot yield a meaningful pretext task. To this end, in this paper we propose a contrastive pre-training model with adversarial perturbations for check-in sequence representation learning (CACSR). Firstly, we design a novel spatial-temporal augmentation block for disturbing the spatial-temporal features of check-in sequences in the latent space to relieve the stress of designing manual data augmentation strategies. Secondly, to construct an effective contrastive pretext task, we generate “hard” positive and negative pairs for the check-in sequence by adversarial training. These two designs encourage the model to capture the high-level spatial-temporal patterns and semantics of check-in sequences while ignoring the noisy and unimportant details. We demonstrate the effectiveness and versatility of CACSR on two kinds of downstream tasks using three real-world datasets. The results show that our model outperforms both the state-of-the-art pre-training methods and the end-to-end models.

Criminal Action Graph: A semantic representation model of judgement documents for legal charge prediction

Semantic information in judgement documents has been an important source in Artificial Intelligence and Law. Sequential representation is the traditional structure for analyzing judgement documents and supporting the legal charge prediction task. The main problem is that it is not effective to represent the criminal semantic information. In this paper, to represent and verify the criminal semantic information such as multi-linked legal features, we propose a novel criminal semantic representation model, which constructs the Criminal Action Graph (CAG) by extracting criminal actions linked in two temporal relationships. Based on the CAG, a Graph Convolutional Network is also adopted as the predictor for legal charge prediction. We evaluate the validity of CAG on the confusing charges which composed of 32,000 judgement documents on five confusing charge sets. The CAG reaches about 88% accuracy averagely, more than 3% over the compared model. The experimental standard deviation also show the stability of our model, which is about 0.0032 on average, nearly 0. The results show the effectiveness of our model for representing and using the semantic information in judgement documents.

Machine Learning Based Prediction of Enzymatic Degradation of Plastics Using Encoded Protein Sequence and Effective Feature Representation

Enzyme biocatalysis for plastic treatment and recycling is an emerging field of growing interest. However, it is challenging and time-consuming to identify plastic-degrading enzymes with desirable functionality, given the large number of putative enzyme sequences. There is a critical need to develop an effective approach to accurately predict the enzyme activity in degrading different types of plastics. In this study, we developed a machine-learning-based plastic enzymatic degradation (PED) framework to predict the ability of an enzyme to degrade plastics of interest by exploring and recognizing hidden patterns in protein sequences. A data set integrating information from a wide range of experimentally verified enzymes and various common plastic substrates was created. A new context-aware enzyme sequence representation (CESR) mechanism was developed to learn the abundant contextual information in enzyme sequences, and feature extraction was performed for enzymes at both the amino acid level and global sequence level. Thirteen machine learning classification algorithms were compared, and XGBoost was identified as the best-performing algorithm. PED achieved an overall accuracy of 90.2% and outperformed sequence-based protein classification models from the existing literature. Furthermore, important enzyme features in plastic degradation were identified and comprehensively interpreted. This study demonstrated a new tool for the prediction and discovery of plastic-degrading enzymes.