Detection of susceptibility to Multiple Sclerosis from Single Nucleotide Polymorphism data
For many diseases that are genetically based, the date of onset is not predetermined or even predictable. To aid in assisting diagnosis of these diseases it is important to understand the person’s susceptibility to developing a particular disease. In this study the susceptibility to Multiple Sclerosis is studied and modelled using as its base SNPs data. A SNP or single nucleotide polymorphism is the name given to a variation is a single base pair in a DNA sequence identified to be at a particular place on a specific chromosome. This data can be obtained using microarray chips which use as their input blood from a sample given by an individual.
To accurately process this data several areas need to be addressed. Firstly, the volume of raw data present, how it is handled, stored and manipulated for later processing. Secondly, how the data can be sensibly reduced to give a more manageable size base from which to model the data, whilst retaining all significant information. Thirdly, the modelling of the data itself and the presentation of these results.. For the purposes of determining susceptibility the modelling draws from the field of classification. To follow this process of investigation the constructivist research approach is followed allowing for results at an earlier stage to alter testing and inference at later stages of investigation. In the first instance a prioritised list of useful methods of data handling, data reduction and modelling can be produced. In the second instance work can then proceed to use these methods to construct a cohesive system of information processing, from the raw data to a “prediction” of susceptibility ideally for a single individual at one time.
These are the aims and processes undertaken in this research, where the desire is to understand the data, its interactions and inter-relations between data points (SNPs), the best processes of data reduction and accurate modelling.
The analysis of experimental data obtained from the Welcome Trust in the UK and results obtained in this study confirm the hypothesis that is it possible to accurately predict the susceptibility of an individual to MS using personalised modelling on SNPs data. His is the first study on this data that results in a high predictive accuracy along with discussing the application of different information methods.
The potential for further work to ensure the methods found here can be implemented into a system usable by clinicians to enhance existing medical procedures is huge. The idea of a personalised model of both the disease and an individual interacting to assist doctors may be closer than previously thought.