Protein Crystallization Segmentation and Classification Using Subordinate Color Channel in Fluorescence Microscopy...

Area of Science:

• Biophysics
• Biochemistry
• Microscopy

Background:

• Accurate protein crystal detection is vital for high-throughput automated systems in drug discovery and structural biology.
• Current trace fluorescent labeling (TFL) methods can lead to misclassifications due to non-specific dye signals.
• Relying solely on peak emission wavelengths for fluorophore analysis is often insufficient for reliable crystal identification.

Purpose of the Study:

• To develop an improved segmentation method for detecting protein crystals in TFL fluorescence microscopy images.
• To enhance the accuracy of automated protein crystallization screening systems.
• To reduce misclassification rates of protein crystals and likely-leads as non-crystals.

Main Methods:

• A novel segmentation method was developed utilizing the percentile intensity of subordinate color channels in emission spectra.
• The method focuses on spectral regions longer than the peak emission wavelength to differentiate true crystal signals.
• Performance was evaluated using trace fluorescently labeled (TFL'd) protein crystallization trial images and validated with new wet lab experiments.

Main Results:

• The subordinate color channel segmentation method significantly reduced misclassification rates, decreasing false negatives from 9.71% to 2.02%.
• Classifier accuracy increased by 1.77% to 5.53% compared to dominant color channel methods.
• The developed method achieved approximately 94% overall accuracy, with misclassification of crystals as non-crystals below 1% in both computational and wet lab tests.

Conclusions:

• Utilizing subordinate color intensity in fluorescence spectra offers a more robust approach for protein crystal detection in TFL images.
• This method enhances the reliability of automated screening systems, reducing errors in identifying potential protein crystals.
• The generalizability of the subordinate channel approach was confirmed through experiments with Concanavalin A and Ab inorganic pyrophosphate.