Scholarlink

<p>Antimicrobial resistance among ESKAPE pathogens, <i>Enterococcus faecium</i>, <i>Staphylococcus aureus</i>, <i>Klebsiella pneumoniae</i>, <i>Acinetobacter baumannii</i>, <i>Pseudomonas aeruginosa</i> and <i>Enterobacter</i> spp. continues to limit therapeutic options. We developed an integrated computational framework for pathogen-specific antimicrobial peptide (AMP) discovery that combines predictive modeling with conditional sequence generation. The predictive modules accurately identified AMPs and assigned pathogen-level activity profiles across ESKAPE organisms. A conditional long short-term memory model (BioAMPify) generated diverse peptides guided by pathogen-specific vectors, preserving key physicochemical properties characteristic of natural AMPs. Generated sequences demonstrated biologically consistent compositional shifts under pathogen conditioning and over 90% were predicted as non-toxic. Compared with alternative generative models, the framework produced peptides with improved safety profiles and minimal overlap with training data. This pathogen-aware strategy integrates activity prediction, sequence design and toxicity screening, providing a computational platform for prioritizing candidate AMPs for experimental validation and therapeutic development against multidrug-resistant pathogens.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Towards Safer Antimicrobial Peptide Therapeutics: A Predictive–Generative Framework Targeting ESKAPE Pathogens

  • 期刊论文

  • 发表日期：

    2026年4月22日

查看全文
  • Zubia Rashid,
  • Hania Ahmed,
  • Kashif Saleem,
  • Muhammad Razi Uddin Siddiqui

摘要

Antimicrobial resistance among ESKAPE pathogens, Enterococcus faecium, Staphylococcus aureus, Klebsiella pneumoniae, Acinetobacter baumannii, Pseudomonas aeruginosa and Enterobacter spp. continues to limit therapeutic options. We developed an integrated computational framework for pathogen-specific antimicrobial peptide (AMP) discovery that combines predictive modeling with conditional sequence generation. The predictive modules accurately identified AMPs and assigned pathogen-level activity profiles across ESKAPE organisms. A conditional long short-term memory model (BioAMPify) generated diverse peptides guided by pathogen-specific vectors, preserving key physicochemical properties characteristic of natural AMPs. Generated sequences demonstrated biologically consistent compositional shifts under pathogen conditioning and over 90% were predicted as non-toxic. Compared with alternative generative models, the framework produced peptides with improved safety profiles and minimal overlap with training data. This pathogen-aware strategy integrates activity prediction, sequence design and toxicity screening, providing a computational platform for prioritizing candidate AMPs for experimental validation and therapeutic development against multidrug-resistant pathogens.