Scholarlink

<p>This study applies Elephant Herding Optimization (EHO) to the Multiple Sequence Alignment (MSA) problem and evaluates its performance against a Genetic Algorithm (GA) under a common objective and budget. EHO operationalizes two herd behaviors: clan-based updating around a matriarch (exploitation) and the separation of adult males (exploration), through the Clan Updating Operator (CUO) and the Separating Operator (SO). This design yields a principled exploration–exploitation balance without disruptive recombination, reducing premature convergence while preserving high-quality partial alignments. Across Alzheimer-related homologous protein targets with 30 runs, EHO consistently attains higher average alignment scores and lower variance than GA, indicating stronger solution quality and more stable convergence. EHO also requires fewer hyperparameters and less operator tuning than GA, simplifying implementation and improving reproducibility. Sensitivity analysis shows that clan size and separation rate effectively modulate intensification and diversification, enabling robust performance on multimodal landscapes where GA is prone to stagnation. Although EHO can incur higher computational costs in naive implementations, its structure is amenable to parallelization and early stopping, mitigating runtime overhead in practice. Overall, EHO provides a robust, parameter-light alternative to GA for MSA, delivering superior alignment quality and stability through biologically grounded population dynamics that maintain diversity without sacrificing intensification. We provide the Python implementation through a web link <a href="https://github.com/riosew/Elephant-Herding-Algorithm-for-the-Multiple-Sequence-Alignment-Problem.git">https://github.com/riosew/Elephant-Herding-Algorithm-for-the-Multiple-Sequence-Alignment-Problem.git</a>.</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

Elephant herding algorithm for the multiple sequence alignment problem

  • 期刊论文

  • 发表日期：

    2026年4月30日

查看全文
  • Ernesto Rios-Willars,
  • María Magdalena Delabra-Salinas,
  • Ricardo Eduardo Carrillo-Gaona

摘要

This study applies Elephant Herding Optimization (EHO) to the Multiple Sequence Alignment (MSA) problem and evaluates its performance against a Genetic Algorithm (GA) under a common objective and budget. EHO operationalizes two herd behaviors: clan-based updating around a matriarch (exploitation) and the separation of adult males (exploration), through the Clan Updating Operator (CUO) and the Separating Operator (SO). This design yields a principled exploration–exploitation balance without disruptive recombination, reducing premature convergence while preserving high-quality partial alignments. Across Alzheimer-related homologous protein targets with 30 runs, EHO consistently attains higher average alignment scores and lower variance than GA, indicating stronger solution quality and more stable convergence. EHO also requires fewer hyperparameters and less operator tuning than GA, simplifying implementation and improving reproducibility. Sensitivity analysis shows that clan size and separation rate effectively modulate intensification and diversification, enabling robust performance on multimodal landscapes where GA is prone to stagnation. Although EHO can incur higher computational costs in naive implementations, its structure is amenable to parallelization and early stopping, mitigating runtime overhead in practice. Overall, EHO provides a robust, parameter-light alternative to GA for MSA, delivering superior alignment quality and stability through biologically grounded population dynamics that maintain diversity without sacrificing intensification. We provide the Python implementation through a web link https://github.com/riosew/Elephant-Herding-Algorithm-for-the-Multiple-Sequence-Alignment-Problem.git.