WormSpot: a machine learning-powered viability scoring platform in C. elegans for Candida pathogenicity studies.

Area of Science:

• Microbiology
• Computational Biology
• Genetics

Background:

• Invasive Candida infections are a growing concern due to rising antifungal resistance.
• Caenorhabditis elegans (C. elegans) is a valuable model for studying Candida pathogenicity, but traditional viability assays are inefficient for high-throughput studies.
• Existing methods for assessing C. elegans survival after infection are labor-intensive and limit scalability.

Purpose of the Study:

• To develop an automated, data-efficient platform for scoring C. elegans viability post-Candida infection.
• To create a machine learning model capable of accurately classifying worm viability from static images.
• To enhance the scalability of C. elegans infection models for virulence assessment and antifungal screening.

Main Methods:

• Developed WormSpot, a machine learning platform utilizing the YOLOv8 framework.
• Trained the model on representative images of Candida-infected C. elegans.
• Validated the platform's performance against manual scoring and known Candida albicans mutants and antifungal agents.

Main Results:

• WormSpot accurately classifies worm viability, showing strong concordance with manual scoring.
• The platform requires minimal image input and performs efficiently with diverse image formats.
• Demonstrated robustness in predicting worm survival trends and compatibility with multi-well microscopy workflows.

Conclusions:

• WormSpot provides a data-efficient and reproducible tool for assessing Candida virulence in C. elegans.
• The platform supports high-throughput screening for antifungal discovery and basic pathogenicity research.
• WormSpot offers a generalizable framework for host-pathogen phenotyping in microbiology.