UniDL4BioPep: a universal deep learning architecture for binary classification in peptide bioactivity.

Identification of potent peptides through model prediction can reduce benchwork in wet experiments. However, the conventional process of model buildings can be complex and time consuming due to challenges such as peptide representation, feature selection, model selection and hyperparameter tuning. Recently, advanced pretrained deep learning-based language models (LMs) have been released for protein sequence embedding and applied to structure and function prediction. Based on these developments, we have developed UniDL4BioPep, a universal deep-learning model architecture for transfer learning in bioactive peptide binary classification modeling. It can directly assist users in training a high-performance deep-learning model with a fixed architecture and achieve cutting-edge performance to meet the demands in efficiently novel bioactive peptide discovery. To the best of our best knowledge, this is the first time that a pretrained biological language model is utilized for peptide embeddings and successfully predicts peptide bioactivities through large-scale evaluations of those peptide embeddings. The model was also validated through uniform manifold approximation and projection analysis. By combining the LM with a convolutional neural network, UniDL4BioPep achieved greater performances than the respective state-of-the-art models for 15 out of 20 different bioactivity dataset prediction tasks. The accuracy, Mathews correlation coefficient and area under the curve were 0.7-7, 1.23-26.7 and 0.3-25.6% higher, respectively. A user-friendly web server of UniDL4BioPep for the tested bioactivities is established and freely accessible at https://nepc2pvmzy.us-east-1.awsapprunner.com. The source codes, datasets and templates of UniDL4BioPep for other bioactivity fitting and prediction tasks are available at https://github.com/dzjxzyd/UniDL4BioPep.

Structured Sparse Regularized TSK Fuzzy System for predicting therapeutic peptides.

Therapeutic peptides act on the skeletal system, digestive system and blood system, have antibacterial properties and help relieve inflammation. In order to reduce the resource consumption of wet experiments for the identification of therapeutic peptides, many computational-based methods have been developed to solve the identification of therapeutic peptides. Due to the insufficiency of traditional machine learning methods in dealing with feature noise. We propose a novel therapeutic peptide identification method called Structured Sparse Regularized Takagi-Sugeno-Kang Fuzzy System on Within-Class Scatter (SSR-TSK-FS-WCS). Our method achieves good performance on multiple therapeutic peptides and UCI datasets.

A novel antibacterial peptide recognition algorithm based on BERT.

As the best substitute for antibiotics, antimicrobial peptides (AMPs) have important research significance. Due to the high cost and difficulty of experimental methods for identifying AMPs, more and more researches are focused on using computational methods to solve this problem. Most of the existing calculation methods can identify AMPs through the sequence itself, but there is still room for improvement in recognition accuracy, and there is a problem that the constructed model cannot be universal in each dataset. The pre-training strategy has been applied to many tasks in natural language processing (NLP) and has achieved gratifying results. It also has great application prospects in the field of AMP recognition and prediction. In this paper, we apply the pre-training strategy to the model training of AMP classifiers and propose a novel recognition algorithm. Our model is constructed based on the BERT model, pre-trained with the protein data from UniProt, and then fine-tuned and evaluated on six AMP datasets with large differences. Our model is superior to the existing methods and achieves the goal of accurate identification of datasets with small sample size. We try different word segmentation methods for peptide chains and prove the influence of pre-training steps and balancing datasets on the recognition effect. We find that pre-training on a large number of diverse AMP data, followed by fine-tuning on new data, is beneficial for capturing both new data's specific features and common features between AMP sequences. Finally, we construct a new AMP dataset, on which we train a general AMP recognition model.

ProtPlat: an efficient pre-training platform for protein classification based on FastText.

BACKGROUND: For the past decades, benefitting from the rapid growth of protein sequence data in public databases, a lot of machine learning methods have been developed to predict physicochemical properties or functions of proteins using amino acid sequence features. However, the prediction performance often suffers from the lack of labeled data. In recent years, pre-training methods have been widely studied to address the small-sample issue in computer vision and natural language processing fields, while specific pre-training techniques for protein sequences are few. RESULTS: In this paper, we propose a pre-training platform for representing protein sequences, called ProtPlat, which uses the Pfam database to train a three-layer neural network, and then uses specific training data from downstream tasks to fine-tune the model. ProtPlat can learn good representations for amino acids, and at the same time achieve efficient classification. We conduct experiments on three protein classification tasks, including the identification of type III secreted effectors, the prediction of subcellular localization, and the recognition of signal peptides. The experimental results show that the pre-training can enhance model performance effectively and ProtPlat is competitive to the state-of-the-art predictors, especially for small datasets. We implement the ProtPlat platform as a web service ( https://compbio.sjtu.edu.cn/protplat ) that is accessible to the public. CONCLUSIONS: To enhance the feature representation of protein amino acid sequences and improve the performance of sequence-based classification tasks, we develop ProtPlat, a general platform for the pre-training of protein sequences, which is featured by a large-scale supervised training based on Pfam database and an efficient learning model, FastText. The experimental results of three downstream classification tasks demonstrate the efficacy of ProtPlat.

Structural classification of proteins using texture descriptors extracted from the cellular automata image.

Nowadays, having knowledge about cellular attributes of proteins has an important role in pharmacy, medical science and molecular biology. These attributes are closely correlated with the function and three-dimensional structure of proteins. Knowledge of protein structural class is used by various methods for better understanding the protein functionality and folding patterns. Computational methods and intelligence systems can have an important role in performing structural classification of proteins. Most of protein sequences are saved in databanks as characters and strings and a numerical representation is essential for applying machine learning methods. In this work, a binary representation of protein sequences is introduced based on reduced amino acids alphabets according to surrounding hydrophobicity index. Many important features which are hidden in these long binary sequences can be clearly displayed through their cellular automata images. The extracted features from these images are used to build a classification model by support vector machine. Comparing to previous studies on the several benchmark datasets, the promising classification rates obtained by tenfold cross-validation imply that the current approach can help in revealing some inherent features deeply hidden in protein sequences and improve the quality of predicting protein structural class.

Predictive Recognition of DNA-binding Proteins Based on Pre-trained Language Model BERT.

Identifying proteins is crucial for disease diagnosis and treatment. With the increase of known proteins, large-scale batch predictions are essential. However, traditional biological experiments being time-consuming and expensive are difficult to accomplish this task efficiently. Nevertheless, deep learning algorithms based on big data analysis have manifested potential in this aspect. In recent years, language representation models, especially BERT, have made significant advancements in natural language processing. In this paper, using three protein segmentation methods and three encoder numbers, nine BERT models with different sizes are constructed to predict whether known proteins are DNA-binding proteins or not. Furthermore, based on the concept of protein motifs, multi-scale convolutional networks are fused into the models to extract the local features of DNA-binding proteins. Finally, we find that the larger the number of encoders, the better the model predictions under the condition of considering each amino acid in the protein as a word. Our proposed algorithm achieves 81.88% sensitivity and 0.39 MCC value on the test set. Furthermore, it achieves 62.41% accuracy on the independent test set PDB2272. It is evident that our proposed method can be a tool to assist in the identification of DNA-binding proteins.

Hamlet-Pattern-Based Automated COVID-19 and Influenza Detection Model Using Protein Sequences.

SARS-CoV-2 and Influenza-A can present similar symptoms. Computer-aided diagnosis can help facilitate screening for the two conditions, and may be especially relevant and useful in the current COVID-19 pandemic because seasonal Influenza-A infection can still occur. We have developed a novel text-based classification model for discriminating between the two conditions using protein sequences of varying lengths. We downloaded viral protein sequences of SARS-CoV-2 and Influenza-A with varying lengths (all 100 or greater) from the NCBI database and randomly selected 16,901 SARS-CoV-2 and 19,523 Influenza-A sequences to form a two-class study dataset. We used a new feature extraction function based on a unique pattern, HamletPat, generated from the text of Shakespeare's Hamlet, and a signum function to extract local binary pattern-like bits from overlapping fixed-length (27) blocks of the protein sequences. The bits were converted to decimal map signals from which histograms were extracted and concatenated to form a final feature vector of length 1280. The iterative Chi-square function selected the 340 most discriminative features to feed to an SVM with a Gaussian kernel for classification. The model attained 99.92% and 99.87% classification accuracy rates using hold-out (75:25 split ratio) and five-fold cross-validations, respectively. The excellent performance of the lightweight, handcrafted HamletPat-based classification model suggests that it can be a valuable tool for screening protein sequences to discriminate between SARS-CoV-2 and Influenza-A infections.

The limitations of the current protein classification tools in identifying lipolytic features in putative bacterial lipase sequences.

Metagenomics sequencing has generated millions of new protein sequences, most of them with unknown functions. A relatively quick first step for function assignment is to use the existing public protein databases and their scanning tools. However, to date these tools are not able to identify all sequence features like conserved motifs or patterns. In this study we evaluated the capability of several protein public databases (e.g., InterPro, PROSITE, ESTHER, pfam, AlphaFold etc) and their scanning tools for identifying lipolytic features in 78 putative cold-adapted bacterial lipase sequences. Novel lipases that can tolerate extreme conditions have great biotechnological importance. We obtained the putative cold-adapted lipolytic sequences from the metagenomic study of anaerobic psychrophilic microbial community treating domestic wastewater at 4 and 15 â„ƒ. Both newer and conventional protein classifiers failed to find lipolytic features for most of the putative lipases. InterProScan predicted lipase family membership for only 18 of the putative lipase sequences. For more than half of them (41 out of 78) InterProScan could not predict any protein family membership, let alone find lipolytic features in them. However, when the Lipase Engineering Database and AlphaFold were used, half of those sequences were classified. Conventional databases like PROSITE could find lipolytic patterns for 9 of the putative lipolytic sequences of which only one was identified by InterProScan as a lipase. Moreover, different scanning tools made different and inconsistent predictions for a certain putative lipase sequence. Even InterProScan, which integrates predictions from 13 protein member databases, did not have a consensus prediction for a certain lipase sequence. Our study shows that there is lack of information in public protein databases about bacterial lipase sequences and this limits their lipolytic feature prediction and biotechnological application. The integration of AlphaFold within the InterPro can improve the lipase identification and classification significantly.

3D representations of amino acids-applications to protein sequence comparison and classification.

The amino acid sequence of a protein is the key to understanding its structure and ultimately its function in the cell. This paper addresses the fundamental issue of encoding amino acids in ways that the representation of such a protein sequence facilitates the decoding of its information content. We show that a feature-based representation in a three-dimensional (3D) space derived from amino acid substitution matrices provides an adequate representation that can be used for direct comparison of protein sequences based on geometry. We measure the performance of such a representation in the context of the protein structural fold prediction problem. We compare the results of classifying different sets of proteins belonging to distinct structural folds against classifications of the same proteins obtained from sequence alone or directly from structural information. We find that sequence alone performs poorly as a structure classifier. We show in contrast that the use of the three dimensional representation of the sequences significantly improves the classification accuracy. We conclude with a discussion of the current limitations of such a representation and with a description of potential improvements.