"SPOTS BEFORE MY EYES" - 2D GELS AND BIOLOGICAL FUNCTIONS
Dr Marketa Zvelebil, Ludwig Institute for Cancer Research, London

Abstract: Discrete elements of the proteomics multi-step process are today well characterised and very robust. In fact many manufacturers boast increasing levels of automation at all stages of the process. Those however who wish to accelerate the productivity of proteomics projects in the future must be able to make intelligent and rapid decisions about their experiment. For example, to reduce unnecessary analysis, highlight poor quality separations and so eliminate poor data and to determine more favourable paths for research quickly. In order to harness the evolving high levels of automation and speed of analysis, in order to make these significant gains in productivity, it is no longer appropriate to take an exclusive `gel view' or `instrument view' of the proteomics experiment. It must be possible to provide any view of the data produced during the current or any similar previous experiment. It must also be possible to provide this data an integrated database environment.

To eliminate direct dependence on particular proprietary data formats and software and so to `open' up `any' data to this independent multiple access requires the selection of a backbone infrastructure coupled with the ability to simply plug-in different data and instrument types. This infrastructure must also provide a modern relational database environment so that entities used in the laboratory can be easily mapped into the database and linked together to provide the complete inter-working of data as well as easy data querying.

Initial efforts are focused on the database configuration for the processing of varied tissue samples with focus on cancer related tissues and for collecting and interpreting gel data from the Melanietm image analysis software. Tools were written, using ORACLE's reports/application forms and Javatm to easily display many-gel-differentials from data within projects and to embed novel sub-cluster and cluster algorithms into reports which aid in the analysis and data mining of 2D gel information. Data mining of this type leads to a number of exciting opportunities and provides a springboard for further analysis of data by mass spec, sequence analysis and bioinformatics techniques within an integrated user-friendly environment.

Examples of the work completed will be presented in addition to data mining and analysis of studies on the cellular responses to concurrent stimulation of receptor tyrosine kinases and the putative implications in treatment and drug design. The advantages of this system, the global vision and future work will be outlined.

This seminar was held at the Department of Computer Science, Royal Holloway, University of London on 15 January 2001.

back