Sequential Properties Representation Scheme for Recurrent Neural Network-Based Prediction of Therapeutic Peptides

Abstract

The discovery of therapeutic peptides is often accelerated by means of virtual screening supported by machine learning-based predictive models. The predictive performance of such models is sensitive to the choice of data and its representation scheme. While the peptide physicochemical and compositional representations fail to distinguish sequence permutations, the amino acid arrangement within the sequence lacks the important information contained in physicochemical, conformational, topological, and geometrical properties. In this paper, we propose a solution to the identified information gap by implementing a hybrid scheme that complements the best traits from both approaches with the aim of predicting antimicrobial and antiviral activities based on experimental data from DRAMP 2.0, AVPdb, and Uniprot data repositories. Using the Friedman test of statistical significance, we compared our hybrid, sequential properties approach to peptide properties, one-hot vector encoding, and word embedding schemes in the 10-fold cross-validation setting, with respect to the F1 score, Matthews correlation coefficient, geometric mean, recall, and precision evaluation metrics. Moreover, the sequence modeling neural network was employed to gain insight into the synergic effect of both properties-and amino acid order-based predictions. The results suggest that sequential properties significantly (P < 0.01) surpasses the aforementioned state-of-the-art representation schemes. This makes it a strong candidate for increasing the predictive power of screening methods based on machine learning, applicable to any category of peptides.

Deep learning-based multi-functional therapeutic peptides prediction with a multi-label focal dice loss function

“…To further demonstrate the power of the ETFC model, we compare it with the existing methods. To improve the reproducibility and reliability of MPMABP ( Li et al 2022 ), MLBP ( Tang et al 2022 ), sequential properties-recurrent neural network (SP-RNN) ( Otović et al 2022 ) and PrMFTP ( Yan et al 2022 ), we provide the hyperparameter details of these models in Supplementary Tables S4–S7 , respectively. The comparison of ETFC with MPMABP, MLBP, SP-RNN, and PrMFTP is performed on the test set, and we randomly selected 80% of the set as the subset and repeated this process five times to obtain five subsets.…”

Antiviral Peptide-Based Conjugates: State of the Art and Future Perspectives

“…A suitable algorithm guiding conjugate design based on activity, serum stability and penetration capability would be desirable to speed up the development process. While algorithms derived from random peptide library design [ 91 ], machine learning-based predictive models [ 92 ] and natural sequence scanning [ 93 ] have been proposed for predicting peptide bioactivity, unfortunately, no such tool is yet available for peptide–antiviral-drug conjugate activity. Future studies should also focus on PDCs with various therapeutic profiles for further development in stability and performance.…”

Specific Focus on Antifungal Peptides against Azole Resistant Aspergillus fumigatus: Current Status, Challenges, and Future Perspectives

“…Furthermore, Capecchi et al [ 134 ] trained RNNs and identified eight non-hemolytic molecules that target different bacteria. Otovic et al [ 135 ] recently used a long-term generative memory RNN to engineer a PEP-137 peptide whose administration enhanced the survival rate to 50% in a murine model of Klebsiella pneumoniae -induced sepsis.…”