Background: Although the development of sequencing technologies has provided a large number of protein sequences, the analysis of functions that each one plays is still difficult due to the efforts of laboratorial methods, making necessary the usage of computational methods to decrease this gap. As the main source of information available about proteins is their sequences, approaches that can use this information, such as classification based on the patterns of the amino acids and the inference based on sequence similarity using alignment tools, are able to predict a large collection of proteins. The methods available in the literature that use this type of feature can achieve good results, however, they present restrictions of protein length as input to their models. In this work, we present a new method, called TEMPROT, based on the fine-tuning and extraction of embeddings from an available architecture pre-trained on protein sequences. We also describe TEMPROT+, an ensemble between TEMPROT and BLASTp, a local alignment tool that analyzes sequence similarity, which improves the results of our former approach. Results: The evaluation of our proposed classifiers with the literature approaches has been conducted on our dataset, which was derived from CAFA3 challenge database. Both TEMPROT and TEMPROT+ achieved competitive results on $$F_{\max }$$F max , $$S_{\min }$$S min , AuPRC and IAuPRC metrics on Biological Process (BP), Cellular Component (CC) and Molecular Function (MF) ontologies compared to state-of-the-art models, with the main results equal to 0.581, 0.692 and 0.662 of $$F_{\max }$$F max on BP, CC and MF, respectively. Conclusions: The comparison with the literature showed that our model presented competitive results compared the state-of-the-art approaches considering the amino acid sequence pattern recognition and homology analysis. Our model also presented improvements related to the input size that the model can use to train compared to the literature methods.

The prediction of secondary protein structures is one of the classic problems of bioinformatics and has several practical applications. In this work, we present an ensemble of bidirectional recurrent networks (capable of doing long-term analyses of amino acid sequences) with random forests (capable of doing local analyses). The fusion was performed using weights for each of the classifiers and classes, found by searching through genetic algorithms. To evaluate our method, we used the PDB, CB6133 and CB513 datasets. We achieved 73.1%, 59.1% and 55.8% Q8 accuracy on PDB, CB6133 and CB513 proteins, respectively, whereas 81.5% Q3 accuracy on PDB proteins using only amino acid sequence information. In order to compare our results against the literature, we also evaluated our method using amino acid sequence and sequence profile features, achieving 73.4% and 68.9% Q8 accuracy on CB6133 and CB513. Our method yielded good results when using only the amino acid sequence and presents competitive results compared to the literature when using amino acid sequence information and protein sequence similarity.

Protein secondary structures are important in many biological processes and applications. Due to advances in sequencing methods, there are many proteins sequenced, but fewer proteins with secondary structures defined by laboratory methods. With the development of computer technology, computational methods have (started to) become the most important methodologies for predicting secondary structures. We evaluated two different approaches to this problem—driven by the recent results obtained by computational methods in this task—(i) template-free classifiers, based on machine learning techniques; and (ii) template-based classifiers, based on searching tools. Both approaches are formed by different sub-classifiers—six for template-free and two for template-based, each with a specific view of the protein. Our results show that these ensembles improve the results of each approach individually.

Sequencing technologies have advanced considerably in recent years, leading to the sequencing of a vast number of proteins through laboratory methods. However, the functional annotation of these proteins has not kept pace with sequencing efforts, creating a significant gap between sequenced proteins and those with known functions. To address this challenge, computational approaches based solely on amino acid sequence features have been developed to improve functional predictions. In this study, we introduce two novel approaches, one based on machine learning and another using an ensemble of machine learning with local alignment. Our machine learning-based model (SUPERMAGOv2) utilizes transformer-based backbones to extract features from multiple layers, which are then processed by six multilayer perceptrons that incorporate a novel bitscore-weighted input derived from DIAMOND alignments, and by an image classification model that converts the extracted feature vectors into images. Furthermore, we present SUPERMAGOv2+, an ensemble model that combines SUPERMAGOv2 with enhanced DIAMOND-based predictions. In addition, we introduce SUPERMAGOv2+Web, a lightweight web server version of SUPERMAGOv2+. Both proposed methods consistently outperform state-of-the-art approaches across various analyses, establishing themselves as leading methodologies for protein function classification based on amino acid sequences.