Anmerkungen:
Diese Datenquelle enthält auch Bestandsnachweise, die nicht zu einem Volltext führen.
Beschreibung:
Peptides are important therapeutic molecules due to their biocompatibility, biodegradability, and selectivity. Their biochemistry makes peptides suitable for mimicking the binding site of proteins, for the inhibition of disease-relevant protein-protein interactions, and to address the problem of multi-drug resistance, among other applications. Therefore, much attention has been devoted in recent years to the design and optimization of bioactive peptides. Frequently, the discovery of new drugs starts with the analysis of large peptide libraries. However, the experimental screening of such libraries is expensive and time-consuming. In silico approaches that potentially reduce the list of candidates for further improvement are essential for modern drug design. Several machine-learning-based predictors of protein-protein interactions have emerged in the last decades. Based on the available information, these predictors have been trained, for instance, to detect protein interactions or the lack of them (classification problem), or to predict binding affinity (BA) as a regression problem. However, regardless of the output variable, most models introduced so far suffer from low generalization capabilities, displaying high variance when predicting unseen data. Additionally, within the context of protein-protein and protein-ligand interactions, most methods contemplate peptides in the same way as proteins or small organic ligands. This consideration underestimates the specificity of short peptide sequences and results in poor performance in predicting protein-peptide interactions. Similarly, machine-learning-based methods aiming to identify therapeutic molecules, such as antimicrobial peptides (AMPs), have been introduced. However, many of these methods are not able to predict a specific function for putative AMPs, such as antibacterial activity. Consequently, in the search for bioactive peptides to address multi-drug resistance in bacteria, state-of-the-art tools display limited precision in predicting antibacterial ...