Deep Learning for Novel Antimicrobial Peptide Design

Abstract

Antimicrobial resistance is an increasing issue in healthcare as the overuse of antibacterial agents rises during the COVID-19 pandemic. The need for new antibiotics is high, while the arsenal of available agents is decreasing, especially for the treatment of infections by Gram-negative bacteria like Escherichia coli. Antimicrobial peptides (AMPs) are offering a promising route for novel antibiotic development and deep learning techniques can be utilised for successful AMP design. In this study, a long short-term memory (LSTM) generative model and a bidirectional LSTM classification model were constructed to design short novel AMP sequences with potential antibacterial activity against E. coli. Two versions of the generative model and six versions of the classification model were trained and optimised using Bayesian hyperparameter optimisation. These models were used to generate sets of short novel sequences that were classified as antimicrobial or non-antimicrobial. The validation accuracies of the classification models were 81.6–88.9% and the novel AMPs were classified as antimicrobial with accuracies of 70.6–91.7%. Predicted three-dimensional conformations of selected short AMPs exhibited the alpha-helical structure with amphipathic surfaces. This demonstrates that LSTMs are effective tools for generating novel AMPs against targeted bacteria and could be utilised in the search for new antibiotics leads.

Discovering highly potent antimicrobial peptides with deep generative model HydrAMP

“…As such, the QSAR methods can only score existing peptides and are not able to directly generate new ones. Another approach is to use autoregressive models trained on AMP sequences for peptide generation [16, 17, 18]. To generate new peptides, these models operate in an iterative manner.…”

Antimicrobial Peptides: An Update on Classifications and Databases

“…A multi-scale convolutional network model (deep neural network) outperformed existing state-of-the-art models when used for AMP discovery [ 546 ], and a long short-term memory (LSTM) generative model and bidirectional LSTM classification model were effective at generating novel antibacterial AMPs that could be utilized as new antibiotic leads [ 547 ]. SVM combined with deep learning-based features identified 436 possible antimicrobial proteins in the genome of Helobdella robusta [ 548 ]. Discriminant analysis (DA), which is a multivariate approach [ 549 ], quadratic discriminate analysis [ 550 ], and conditional random fields [ 551 ] may also be used for AMP prediction.…”

Machine Learning in Antibacterial Drug Design

“…The latter can serve as active agents, starting points for the design of peptidomimetics, or probes for further studies. The field and in silico tools have been reviewed previously ( Lee et al, 2017 ; Cardoso et al, 2020 ; Wang et al, 2021 ), with the emphasis on machine learning–enabled antimicrobial peptide discovery and SVM for the discovery of membrane-active peptides ( Lee et al, 2018 ). However, Frecer reported a successful design of cationic antibacterial peptides derived from protegrin-1 as early as 2006 ( Frecer, 2006 ), and machine learning methodology contributed significantly to the design and discovery of novel peptides, as demonstrated by Fjell et al To single out just one report, they reinforced the traditional QSAR approach with an artificial neural network model (ANN) that inferred a set of peptides with known antibacterial properties from computed descriptors (MOE software).…”