By: Lilian G.
Year: 2024
School: University High
Grade: 11
Science Teacher: Tim Smay
Liquid-liquid phase separation (LLPS), a process where biomolecular condensates form from proteins or nucleic acids, plays a critical role in various cellular functions by housing diverse chemical reactions. However, the dysregulation of LLPS has been linked to severe health conditions, including tumorigenesis and neurodegenerative diseases. This underscores the necessity for accurate and reproducible quantification methods, a challenge that Lilian’s project aims to address.
Liquid-liquid phase separation is essential for the formation of biomolecular condensates, which are membraneless droplets that compartmentalize cellular reactions. These condensates are involved in crucial biological processes like chromatin organization, DNA repair, and transcription. Given the critical roles they play, any dysregulation can lead to significant health issues, making it vital to quantify the frequency, size, shape, and specific roles of these condensates accurately.
Current methods for quantifying LLPS vary widely, leading to inconsistencies and challenges in reproducibility. To overcome these obstacles, Lilian’s project developed a standardized program to quantify condensates more reliably. The program leverages a combination of Python in Spyder, Cellpose, ImageJ, and specific images where LLPS can be observed, streamlining the process into a reproducible and efficient workflow.
The Engineering Approach
Lilian’s program utilizes a multi-step process to ensure accurate quantification:
- Image Acquisition: The first step involves obtaining images where LLPS can be observed.
- Cell Segmentation: The cells within these images are segmented to isolate the areas of interest.
- Condensate Identification: A Gaussian and Laplacian filter is applied to enhance the visibility of the condensates.
- Condensate Counting: The program analyzes maximum intensity to count the number of condensates brought out by the filters.
This entire process runs efficiently, processing 60 images in about 1.5 minutes, demonstrating both speed and accuracy.
The results from Lilian’s program have shown an overall constant trend, which aligns with expectations for cells with minimal disturbances. This constancy validates the program’s accuracy under controlled conditions. However, variations were observed when external factors, such as the addition of salt, were introduced, further confirming the program’s sensitivity and reliability in detecting changes in condensate formation.
While the program has proven successful as a preliminary step, Lilian identifies several areas for improvement. The program currently struggles with images containing cells of varying intensities, necessitating further refinement. Future enhancements could involve incorporating more advanced machine learning techniques and better filters to improve accuracy.
Additionally, ground truth datasets obtained through high-resolution microscopy will be essential to quantitatively define the presence of condensates, enabling further validation and refinement of the program.