Scholarlink

<p>Environmental fungal pathogens relevant to human and animal health pose significant risks, particularly in regions with intensive farming and climate variability. Several taxa detected in Thailand’s clinical and environmental samples, such as <i>Candida tropicalis</i>, <i>Talaromyces marneffei</i>, and <i>Mucor</i> spp., are listed in the World Health Organization (WHO) Fungal Priority Pathogen List (FPPL). This study integrated next-generation sequencing (NGS) metabarcoding and decision tree models to characterize fungal communities and identify environmental conditions associated with pathogen absence across 18 provinces of Northeast Thailand. Soil samples (<i>n</i> = 121) from rice, cassava, sugarcane, and rubber tree fields were collected in 2022 and analyzed for eight environmental parameters: drought level, soil water content, organic matter, nitrogen, phosphorus, potassium, soil temperature, and soil pH. Decision tree models were trained on these samples to derive absence conditions for nine WHO FPPL taxa, which were validated using an independent test dataset (<i>n</i> = 12) collected in 2025. Absence conditions for <i>Falciformispora senegalensis</i>, <i>Mucor</i> spp., and <i>Talaromyces marneffei</i> achieved perfect precision and recall in the test dataset. Precision is the proportion of samples predicted as absent that are truly absent, and recall is the proportion of truly absent samples correctly identified by the condition. <i>Candida tropicalis</i>, <i>Curvularia lunata</i>, and <i>Lichtheimia</i> spp. showed perfect precision but moderate recall (0.42–0.75). Conditions for <i>Scedosporium</i> spp. and <i>Acremonium</i> spp. did not generalize due to limited representation in the training data. Over three years, the fungal community became less diverse and more taxonomically consolidated, coinciding with drought intensification and nutrient shifts. Overall, combining metabarcoding with decision tree models provides a practical framework for identifying low-risk soils and supporting agricultural practices and public health surveillance.</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

Predicting the absence of World Health Organization fungal priority pathogens in agricultural soils using machine learning and ITS metabarcoding

  • 期刊论文

  • 发表日期：

    2026年7月21日

查看全文
  • Syahriar Nur Maulana Malik Ibrahim,
  • Chayawan Jaikla,
  • Nuttapon Pombubpa

摘要

Environmental fungal pathogens relevant to human and animal health pose significant risks, particularly in regions with intensive farming and climate variability. Several taxa detected in Thailand’s clinical and environmental samples, such as Candida tropicalis, Talaromyces marneffei, and Mucor spp., are listed in the World Health Organization (WHO) Fungal Priority Pathogen List (FPPL). This study integrated next-generation sequencing (NGS) metabarcoding and decision tree models to characterize fungal communities and identify environmental conditions associated with pathogen absence across 18 provinces of Northeast Thailand. Soil samples (n = 121) from rice, cassava, sugarcane, and rubber tree fields were collected in 2022 and analyzed for eight environmental parameters: drought level, soil water content, organic matter, nitrogen, phosphorus, potassium, soil temperature, and soil pH. Decision tree models were trained on these samples to derive absence conditions for nine WHO FPPL taxa, which were validated using an independent test dataset (n = 12) collected in 2025. Absence conditions for Falciformispora senegalensis, Mucor spp., and Talaromyces marneffei achieved perfect precision and recall in the test dataset. Precision is the proportion of samples predicted as absent that are truly absent, and recall is the proportion of truly absent samples correctly identified by the condition. Candida tropicalis, Curvularia lunata, and Lichtheimia spp. showed perfect precision but moderate recall (0.42–0.75). Conditions for Scedosporium spp. and Acremonium spp. did not generalize due to limited representation in the training data. Over three years, the fungal community became less diverse and more taxonomically consolidated, coinciding with drought intensification and nutrient shifts. Overall, combining metabarcoding with decision tree models provides a practical framework for identifying low-risk soils and supporting agricultural practices and public health surveillance.