Machine Learning Guided Discovery of Non‐Hemolytic Membrane Disruptive Anticancer Peptides

Abstract

Most antimicrobial peptides (AMPs) and anticancer peptides (ACPs) fold into membrane disruptive cationic amphiphilic αhelices, many of which are however also unpredictably hemolytic and toxic. Here we exploited the ability of recurrent neural networks (RNN) to distinguish active from inactive and non-hemolytic from hemolytic AMPs and ACPs to discover new non-hemolytic ACPs. Our discovery pipeline involved: 1) sequence generation using either a generative RNN or a genetic algorithm, 2) RNN classification for activity and hemolysis, 3) selection for sequence novelty, helicity and amphiphilicity, and 4) synthesis and testing. Experimental evaluation of thirty-three peptides resulted in eleven active ACPs, four of which were non-hemolytic, with properties resembling those of the natural ACP lasioglossin III. These experiments show the first example of direct machine learning guided discovery of non-hemolytic ACPs.

To Fold or Not to Fold: Diastereomeric Optimization of an α-Helical Antimicrobial Peptide

“…Furthermore, diastereomers were generally toxic against A549 lung cancer cells, with HP11 showing the strongest toxicity (IC 50 = 3.6 ± 0.1 μM), in line with the fact that many AMPs are often active against cancer cells (Figures S5 and S6). , The observed differences between diastereomers in hemolysis, toxicity against HEK293 cells or A549 lung cancer cells, are probably caused by diastereomeric interactions with the different membrane components of the different cell types and possibly proteins in the cell culture medium …”

Efficient Computational Framework for Target-Specific Active Peptide Discovery: A Case Study on IL-17C Targeting Cyclic Peptides

“…Recently, some AI-based predictive models have also been developed to predict potential bioactive peptides. 25,26 Conventional prediction models rely on ML algorithms and feature abstraction techniques, such as compositional features, positional features, and physicochemical properties, to establish mapping relationships between sequences and properties. 27 For instance, Tyagi et al developed a platform based on the support vector machine for identifying novel anticancer peptides (ACPs), 28 while Plisson et al integrated ML algorithms with an outlier detection strategy to identify antimicrobial peptides (AMPs) with nonhemolytic properties.…”

“…Despite the increasing research into the automated design of unique molecules desired by researchers, the conditional control of the characteristics of the generated molecules remains a nontrivial task, such as the generation of peptide molecules with high binding activity to specific target proteins. Recently, some AI-based predictive models have also been developed to predict potential bioactive peptides. , Conventional prediction models rely on ML algorithms and feature abstraction techniques, such as compositional features, positional features, and physicochemical properties, to establish mapping relationships between sequences and properties . For instance, Tyagi et al developed a platform based on the support vector machine for identifying novel anticancer peptides (ACPs), while Plisson et al integrated ML algorithms with an outlier detection strategy to identify antimicrobial peptides (AMPs) with nonhemolytic properties .…”

Multi-dimensional deep learning drives efficient discovery of novel neuroprotective peptides from walnut protein isolates

“…21 RNN was used to discover new non-hemolytic anticancer peptides. 22 Zhang et al 23 compared the classification performance of four natural language processing neural network models, including LSTM, CNN and bidirectional encoder representation transformer (BERT_base and ProtBERT), and the best performing one was used for screening of antihypertensive peptides. These showed that deep learning strategies have been extensively applied to the screening of active peptides with state-of-the-art performance, which reduced the time and errors of complex separation and purification steps.…”