Scholarlink

<p>Metaheuristic applications in bioinformatics are limited, while significant advancements in classification and recognition systems will aid medical experts in the diagnosis of disease for a healthy life. This study aimed to predict the location of the proteins within cells, which is a crucial task in computational biology and bioinformatics. Accurate localization is essential to understanding protein function, cellular roles, and disease connections. This study introduces a hybrid metaheuristic approach that combines <b>Scatter Search (SS) with Simulated Annealing (SA)</b> through established probability distributions to predict protein localization sites in Yeast and E. coli datasets. Building upon the Scatter Search-Simulated Annealing Algorithm’s proven success in addressing Flexible Manufacturing System Layout challenges(Krishnan et&#xa0;al. <CitationRef CitationID="CR29">2012</CitationRef>), we present the first application of this hybrid methodology to bioinformatics, specifically targeting protein subcellular localization prediction problem. The approach was evaluated against popular machine learning and deep learning models. Therefore, we compared the proposed method using evaluation metrics with those reported in previous studies. The approach achieved effective results in terms of evaluation metrics with an accuracy <b>(92.54%)</b> and F1-score <b>(92.81%).</b> This research improves our understanding of cellular behavior by using a hybrid metaheuristic technique to predict protein localization in microorganisms, such as Yeast and E. coli. Furthermore, the architecture of the hybrid method was meticulously designed to harness the strengths of both Scatter Search and Simulated Annealing, allowing for an effective exploration of the solution space while avoiding local optima. The integration of these metaheuristic methods presents a stronger optimization process, leading to improved model performance compared to other models.</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

Combining scatter search with simulated annealing for optimizing protein localization prediction

  • 期刊论文

  • 发表日期：

    2026年3月30日

查看全文
  • Eslam E. M. Attia,
  • Osman Ali Sadek Ibrahim,
  • Abdelmgeid A. Ali

摘要

Metaheuristic applications in bioinformatics are limited, while significant advancements in classification and recognition systems will aid medical experts in the diagnosis of disease for a healthy life. This study aimed to predict the location of the proteins within cells, which is a crucial task in computational biology and bioinformatics. Accurate localization is essential to understanding protein function, cellular roles, and disease connections. This study introduces a hybrid metaheuristic approach that combines Scatter Search (SS) with Simulated Annealing (SA) through established probability distributions to predict protein localization sites in Yeast and E. coli datasets. Building upon the Scatter Search-Simulated Annealing Algorithm’s proven success in addressing Flexible Manufacturing System Layout challenges(Krishnan et al. 2012), we present the first application of this hybrid methodology to bioinformatics, specifically targeting protein subcellular localization prediction problem. The approach was evaluated against popular machine learning and deep learning models. Therefore, we compared the proposed method using evaluation metrics with those reported in previous studies. The approach achieved effective results in terms of evaluation metrics with an accuracy (92.54%) and F1-score (92.81%). This research improves our understanding of cellular behavior by using a hybrid metaheuristic technique to predict protein localization in microorganisms, such as Yeast and E. coli. Furthermore, the architecture of the hybrid method was meticulously designed to harness the strengths of both Scatter Search and Simulated Annealing, allowing for an effective exploration of the solution space while avoiding local optima. The integration of these metaheuristic methods presents a stronger optimization process, leading to improved model performance compared to other models.