petiteFinder

petiteFinder automates quantification of petite colony frequencies in Saccharomyces cerevisiae to assess mitochondrial respiration integrity.

Key Features:

• Automated Detection: Uses deep learning and computer vision to detect and differentiate Grande (wild-type) and petite colonies in scanned images of Petri dishes.
• Classification by Size and Morphology: Identifies and classifies colonies based on size and morphology to distinguish Grande and petite phenotypes.
• Small-Object Detection: Employs a model trained to overcome challenges associated with small object detection in existing architectures.
• Throughput and Accuracy: Processes images up to 100× faster than manual counting while maintaining accuracy comparable to human annotation.
• Reproducibility: Automation reduces manual variability to improve reproducibility of petite frequency measurements.
• Scalability: Facilitates large-scale experimental setups for high-throughput assays.
• Experimental Standardization: Provided experimental protocols standardize the assay across research settings.

Scientific Applications:

• Mitochondrial respiration assessment: Quantifies petite colony frequency as a visual marker of non-respiratory capability and mitochondrial function.
• Yeast genetics and mitochondrial biology: Enables studies in Saccharomyces cerevisiae, which tolerates mitochondrial dysfunction under fermentation conditions.
• Phenotype-based screening: Supports screening for mutations or conditions that alter respiratory competence by distinguishing Grande and petite colonies.
• High-throughput quantification: Enables efficient measurement of petite frequencies in large-scale experiments.

Methodology:

Applies computer vision to scanned images of yeast cultures on Petri dishes and uses a deep learning model trained to classify colonies by size and morphology, explicitly addressing small-object detection challenges.