Award details

AMCB: A software toolkit for construction and analysis of Aggregate Maps in Cell Biology

Reference	BB/H013423/1
Principal Investigator / Supervisor	Professor Niall Michael Adams
Co-Investigators / Co-Supervisors	Professor Paul Freemont
Institution	Imperial College London
Department	Mathematics
Funding type	Research
Value (£)	226,681
Status	Completed
Type	Research Grant
Start date	01/10/2010
End date	31/03/2013
Duration	30 months

Abstract

Much analysis of architecture in cell biology use relative assessment of spatial preference. This is usually instantiated by exploring distances between objects of different type. While this is adequate for reasoning about specific objects it does not provide the opportunity to reason about the absolute spatial preference of objects, or how this is affected by experimental perturbation. We hypothesis that there are global principles of architecture in operation, that contribute to the absolute positioning of functional objects. This research will develop a tool called an Aggregate Map (AM) to capture the absolute spatial positioning of functional objects from replicate images. The key technical aspect of the proposal is the development of a methodology for constructing the AM from replicate images. Broadly, the methodology consists of algorithms to do three things: 1) locate landmarks on the nuclear boundary for registration (that is, putting the cells in a common coordinate system); 2) estimation of an average boundary shape, to serve as the common coordinate system; 3) deformation of each cell boundary and its constituents into the common coordinate system. The common coordinate system and the constituents from each cell form the AM. The instantiation of these steps for 3-dimensional images requires methodological research. For (1), the number and position of the landmarks will be explored, and algorithms for optimal selection developed. There are a number of possibilities for (2), including generalised Procrustes analysis. We will compare procedures to determine the best combination of algorithms. Our preference for (3) is thin-plate spines, but we will consider other approaches. Having determined the best way to construct the AM, we will assess its utility and the merits of subsequent analysis procedure, using extensive simulation and real data. We will be especially concerned to verify that our results are consistent with relative analysis methods.

Summary

Nuclear biology is concerned with understanding the construction and operation of the cell nucleus, a major component of mammalian cells, which carries the bulk of the genetic material and regulates many aspects of cellular operation.The cell nucleus is an incredibly complex entity which undergo dynamic changes throughout the cell's lifecycle. Research in cell biology is crucial to fully understand the basic operation of this fundamental unit of life. In turn, this understanding translates into insight about a host of diseases, and the distant promise of diagnostic markers and even cures. In addition to genetic material, the nucleus contains a great deal of other molecular machinery that is crucial for normal operation. Examining suitably prepared microscopy images of the nucleus reveals a complex arrangement of functional objects that instantiate this machinery. These objects, often punctate, appear to scatter randomly inside nucleus. However, close inspection of multiple nuclei and quantitative analysis suggest that the scattering of these objects, while stochastic, appears to follow subtle rules. These observations provide that tantalising suggestion that the spatial layout of the components of the nucleus are, in part, positioned in a manner that reflects their functional roles. However, existing analysis tools are typically based on a relative assessment of spatial preference. For example, functional objects of one type are observed in close proximity to functional objects of a different type. Such analysis is adequate for making statements about the relationship between the specific functional objects, but provides no information about the absolute spatial preference of the objects within the nucleus. This research will provide new data analysis methodology that will explicitly address this shortcoming. The fundamental hypothesis is that there is some absolute spatial organisation that can be captured by examining multiple nuclear images in a novel way. This novel approach consists of a collection of mathematical transformations from the area of statistical shape analysis. These transformations first find the average shape of the nucleus from multiple images, and then use a special mathematical deformation to force each nucleus and it interior into the average shape. In this way, all nuclei are placed in a common coordinate system, which means analysis of absolute spatial preference is possible. The collection of data in the common coordinate system is called an aggregate map (AM). Analysis and visualisation of the AM provides completely new insights about spatial architecture, and offers great opportunities for analysis of perturbed cells. A number of research steps are required to develop this methodology for analysis of 3-dimensional confocal microscopy images, including a very thorough evaluation phase. On completion of this research, we will release AMCB, a software tool for constructing and analysing aggregate maps. This will allow cell biology researchers worldwide to move towards the ultimate end point of a fully functioning model of a virtual cell nucleus and to uncover the basic principles underlying its functional organisation.

Impact Summary

Direct beneficiaries of the AMCB toolkit will be cell biologists, who will benefit from sophisticated analysis tools presented in easy to use software. These tools will enable researchers to answer fundamentally different questions about nuclear architecture. In this way, AMCB has great potential to progress basic understanding of nuclear architecture. This is a low risk activity for researchers, since it complements existing analysis procedures, and provides added value to existing microscopy images. The ability to get more from existing images is an important way of maximising the return from nation's investment in research. The tools provide the means to reason differently about diagnostic markers, and this offers the opportunity to collaborate with pharmaceutical companies. Thinking differently, these tools can be used to assist microscope operators, and this provides a possible link to work with microscopy software manufacturers. Both of these aspects point to the possibility of AMCB being pulled through to benefit a wider community - in the former case by contributing to advanced medical diagnostics and thereby enhancing quality of life. The website supporting AMCB is a key aspect of our dissemination plans. This site will have a technical side, intended to provide a resource to researchers, and a public side intended for a lay audience. The technical side will provide: an alternative means for researchers to access the AMCB tools, the results of AMCB analysis of our large (and growing) database of mammalian cell images, and results of other researchers AMCB analyses. This part of the site is really intended to provide a central repository, to enhance the quantitative and object analysis of nuclear images. Equally important is the 'public understanding' aspect of the site. The objective here is to enable a lay audience to understand some of the sophisticated insights that AMCB is revealing, and their importance, without unnecessary technical distractions. We are collaborating with a statistical colleague at McGill University, Canada, who has contacts with cell biology labs there. This link provides the opportunity for us to develop a broader scientific collaboration. Morever, we will seek other scientific collaborators to be involved in the developing and testing of AMCB, to ensure that it is fit for purpose. The interdisciplinary aspects of this project (biology, statistics, computing) and collaborative aspects of the project will provide the RA with extensive experience that will be applicable in many many sectors.

Committee	Research Committee C (Genes, development and STEM approaches to biology)
Research Topics	Technology and Methods Development
Research Priority	Technology Development for the Biosciences
Research Initiative	X - not in an Initiative
Funding Scheme	X – not Funded via a specific Funding Scheme

I accept the terms and conditions of use (opens in new window)

export PDF file

back to list new search