Creating new varieties that go further in meeting the needs of farmers and consumers in general requires considerable resources for the plant improvement process. The recent development of techniques shedding light on the living organism and the mass of information from different research programmes certainly makes it easier to create new varieties, but also requires better management and processing of data, so as to optimise the use of resources earmarked for research on variety creation.

Variety creation programmes produce large amounts of data, and the widespread use of molecular marking and the results of genome programmes will only increase the mass of information to be processed even more. The main obstacle to effective processing of this data is the diversity of formats, files and databases in which the information is stored. At present, the selectors and other researchers are obliged to consult several files and databases to access all the information they need to make decisions. They each use their own information storage format to gather all of the data they need more or less manually.

With a view to integrating all the data from our research more effectively and to increasing the capacity to process this information better, we are constructing a decisional database, populated by more or less independent operational systems and making multidisciplinary investigations and analyses possible. This central multi-species database will contain all of the useful data for research programmes on the subject of variety creation: marker and mapping data, genotyping data (results), phenotype data, pedigree data, analysis results (QTL and other). Data processing software will be constructed and set up gradually to be used in conjunction with this database. They will be integrated into a data visualisation and analysis platform which, as a whole, will enable the data produced by all LVH research to be exploited to the full.

This will be an information system for exclusive use, hosted on our own servers given the highly confidential nature of the information contained in the database.

Various plant improvement problems cannot be resolved by traditional quantitative genetic theory, as they require complex genetic systems to follow often very restrictive assumptions of the usual models. Those problems that cannot be resolved through analysis can nevertheless be subject to computer simulations. By setting a few genetic and environmental parameters, simulated data amounts can swiftly be generated in silico, which can then be analysed statistically to study correlations, detect trends and assess different strategies of the variety creation process.

We are hoping to develop a flexible and powerful genetic simulation program that will be adaptable to each of the different cases encountered in practice. It must be able to set accurate genetic and environmental parameters, and easily assess the effectiveness of different selection strategies at developing new varieties ever more suited to the target environment. As this type of simulation can consume considerable computer resources in terms of computing power, and as it is a process that is easily parallelisable, this software program is ideal for computing grids. The use of resources offered by Auvergrid is therefore being considered.

Although the provision of part or all of the software programs developed, for organisations who are not competing directly with LVH, could be considered under special circumstances, we would like use of these programs to be controlled.

We are hoping that this project (the database and software programs set up) will result in the following :

• easier access to all relevant information;
• effective processing of all data available to variety creation by selectors and other LVH scientists;
• a significant improvement in the decision-making process on research conducted by LVH researchers;
• possibility of speeding up the creation of varieties that go further in meeting the farming needs of tomorrow.

LifeGrid, the regional information system