Scholarlink

<p>Anthropogenic climate change has had devastating effects on the Florida and Caribbean reef systems, in part due to increased disease outbreaks. Climate change exacerbates marine diseases by expanding pathogen ranges and heightening host susceptibility through environmental stress. Specifically, there has been a stark rise in marine disease events outbreaks targeting multiple coral species, resulting in high mortality rates and declining reef biodiversity. Although many of these diseases present similar visual symptoms, they exhibit varying mortality rates and require distinct treatment protocols. Advances in coral transcriptomics research have enhanced our understanding of coral responses to various diseases, but more sophisticated methods are required to classify diseases that appear visually similar. This study provides the first machine learning (ML) model that can classify two common coral diseases: stony coral tissue loss disease (SCTLD) and white plague (WP). Using various algorithms, 463 gene expression biomarkers were identified, with 275 unique to SCTLD and 167 unique to WP, revealing distinct immune responses between the two diseases. The final ML model was built with partial least squares discriminant analysis (PLS-DA) and the identified biomarkers were tested and validated with samples collected in situ. It achieved high predictive performance, with an Area Under the Receiver Operating Characteristic (ROC) Curve (AUC) of 0.9895, an average overall error rate of 0.0754, and an average balanced error rate (BER) of 0.0799. This study provides a preliminary disease classification model that reliably distinguishes between SCTLD and WP and offers valuable insights into their underlying cellular responses. Additionally, the identified biomarkers provide a foundation for the development of rapid diagnostic tools to identify and mitigate future coral disease outbreaks.</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

Leveraging machine learning to classify and characterize gene expression patterns in two coral diseases

  • 期刊论文

  • 出版模式：

    开放获取

  • 发表日期：

    2025年8月27日

查看全文
  • Emily W. Van Buren,
  • Kelsey M. Beavers,
  • Mariah N. Cornelio,
  • Alexia Stokes,
  • Madison Emery,
  • Marilyn E. Brandt,
  • Jeffery P. Demuth,
  • Li Wang,
  • Laura D. Mydlarz

摘要

Anthropogenic climate change has had devastating effects on the Florida and Caribbean reef systems, in part due to increased disease outbreaks. Climate change exacerbates marine diseases by expanding pathogen ranges and heightening host susceptibility through environmental stress. Specifically, there has been a stark rise in marine disease events outbreaks targeting multiple coral species, resulting in high mortality rates and declining reef biodiversity. Although many of these diseases present similar visual symptoms, they exhibit varying mortality rates and require distinct treatment protocols. Advances in coral transcriptomics research have enhanced our understanding of coral responses to various diseases, but more sophisticated methods are required to classify diseases that appear visually similar. This study provides the first machine learning (ML) model that can classify two common coral diseases: stony coral tissue loss disease (SCTLD) and white plague (WP). Using various algorithms, 463 gene expression biomarkers were identified, with 275 unique to SCTLD and 167 unique to WP, revealing distinct immune responses between the two diseases. The final ML model was built with partial least squares discriminant analysis (PLS-DA) and the identified biomarkers were tested and validated with samples collected in situ. It achieved high predictive performance, with an Area Under the Receiver Operating Characteristic (ROC) Curve (AUC) of 0.9895, an average overall error rate of 0.0754, and an average balanced error rate (BER) of 0.0799. This study provides a preliminary disease classification model that reliably distinguishes between SCTLD and WP and offers valuable insights into their underlying cellular responses. Additionally, the identified biomarkers provide a foundation for the development of rapid diagnostic tools to identify and mitigate future coral disease outbreaks.