Computational model of interphase chromosome territories in a simulated cell nucleus

Project Title: Computational model of interphase chromosome territories in a simulated cell nucleus
Program:
Northeast (308)

Project Leader
--------------
Project Leader:
Andre Khalil

Email: andre.khalil@maine.edu
Mobile Phone: {Empty}
Work Phone: {Empty}

Project Information
-------------------
Project Description:
This project started in the mid 2000s during our work centered on the development of a novel image segmentation approach called the Wavelet Transform Modulus Maxima (WTMM) segmentation method. Our first application on the segmentation and analysis of fluorescence microscopy images of chromosome territories (CTs) from interphase mouse lymphocytes (Khalil et al. 2007 DOI:doi.org/10.1007/s10577-007-1172-8). As an accompaniment to the segmentation approach, the biological analysis motivated us to develop a computational (in silico) random model to which the empirical observations could be compared to. In this model, CTs are modeled as ellipsoids having similar geometrical attributes as the empirical data (in terms of volume, diameter, and surface area), except that they are randomly rotated and randomly positioned within a simulated cell nucleus that is also based on empirical observations. For each realization, measurements such as inter-CT distance, radial distance, relationship between CT size and eccentricity vs. their position within the nucleus, etc., are made and compared to the empirical data obtained from the fluorescence microscopy images. The overall purpose of this computational model is to provide a randomized numerical version of the empirical observations of chromosome territories, to help demonstrate and quantify their non-random (preferential) position. The biomedical context is that in a disease state (e.g. cancer), the shape and position of some CTs are known to be altered, which motivates the development of potential early diagnosis computational approaches.
 
In our most recent work (https://digitalcommons.library.umaine.edu/etd/3416) we used the numerical model to show that a subset of CTs from mouse lymphocytes were closer to the periphery of the cell nucleus when compared random positioning. However, due to computational constraints (both in terms of coding efficiency and computing power), the work was limited in two ways: only one measurement was made (the radial distance) and only one CT was simulated at a time. The computing time for the program to converge when placing several CTs increases greatly with increasing number of CTs. Placing at least four CTs is our goal, so that we can also include inter-CT measurements.