 Hello, my name is Marco Panier and I would like to present you today a meta-netics MNX ref, which is a unified name space for metabolic models. In biology, all biologists have a background in biochemistry and have some notion of an idealized view of all possible biochemical reactions and pathways. For example, for a given organism, we know that a subset of all these reactions might happen in a given organism. From a bioinformatic and chemoinformatic perspective, there is nevertheless a big challenge. It's a database challenge that consists in the problem of knowing how these metabolites and reactions need to be represented for a human and for a computer. Essentially, the goal is to have all these properly connected for a single organism or across all organisms. The global picture is really going through a thorough layer. We have the description of the metabolites, which are related to chemical compounds described in terms of atoms. There is the expression of the biochemical reaction that these are equations with symbols with metabolites as symbols. The global picture is a genome-scale metabolic network for a given organism, which is the basis for a metabolic model. Different databases have attempted to capture and describe these words, and certainly the most famous of them is the Kyoto Anticlopedia of Genes and Genome. More recently, over-specialized databases have appeared to document in more detail the metabolite level. For example, for Kebii or the biochemical reaction for Rea, these two databases are being developed in things like one to the other. But Kebii and Rea do not cover the genome-scale part of the problem. Another resource, the big models, is certainly historically the most important and the major contributor to metabolic models. But also, the big models have been expressed in terms of biochemical equation, and it's only very recently that the chemical compounds have started to be taken into account in the big models. Overdatabase, like Kebii's or Metasike, are attempts to cover the whole range, but this happens at the cost of some idiosyncratic description of this biochemistry. We can have a linguistic analogy here with all these different resources. The problem is the metabolites are expressed using their own identifiers in every database. The chemistry might be slightly different, and already at the level of metabolites, there is some difficulties to build links between the different databases. With the biochemical equation, the situation becomes more complicated, and if we go to the other level of the genome-scale metabolic network, we end up in a situation where models published by different groups are very hard to compare and to reconcile together. And here, exactly, we can introduce mathematics, MNXF, which is primarily a multilingual dictionary for metabolite and reaction to link the major public resources related to metabolism. Very basically, for example, if you have two reactions in two databases, MNXF is built first by attempting to establish identities between metabolites, which help to deduce that the reaction in the two databases are in fact the same reaction. What made the exercise more challenging is the fact that we are willing to deal with genome-scale metabolic network. These are models, and here is a very, very simple example where you can see the metabolism of an organism with the arrow representing the different biochemical reaction. You have some enzyme reaction, you have transport reaction, you have a definition of the growth medium for this organism, and you have the product of growth, the most important one here being the biomass. In addition to the description of the chemistry, we can place on such a model the protein that are responsible for catalyzing the biochemical equation and the transport between the different subcellular compartments. This cartoon network is very simple. I just put here on this slide two examples of models that have been published, and the take-home message from this slide is simply that realistic models have most of a statistic by two or three orders of magnitude larger than the cartoon network. Simply to say that genome-scale metabolic network are complex objects. There exists a range of algorithm that permits to simulate this model. And for example, a typical example is flux balance analysis that allows you to answer questions like, can biomass be produced given the following growth medium? And the algorithm in that case will try to produce biomass, and in this example you see it succeed. Starting from this basis, we can start to ask questions about the different components of the model. For example, we can remove a reaction, run the simulation, say that biomass can still be produced, and we'll reach the conclusion that this reaction is not essential. We do it again with another reaction, and in that case we see that the model stopped functioning. We did use that the reaction is essential, and such predictions are really important because they can be tested against experiments. These models are really unique, worth collecting, regularly published for different organisms, but they are also fragile. Typically 10 or 20% of reaction are essential in this model, which means the removal of any of them kills the models. So grid tar must be taken while manipulating, modifying the model, which is really the major constraint to meet during metabolite reconciliation. Here is just a few numbers to show the different data sources that appear in the latest reconciliation. And you see the total number of metabolite is something like 1 million and about 10,000 of them are retrieved in model. And also we have slightly more than 10,000 reactions that appear in models. This full data set is distributed in top delimited and RDF total formats under CC by license, and it is certainly the way methanetics MNX ref is most used by groups downloading the full data set for their own needs. A few examples of third party application. First are the cross-link between the different resources that are now exposed in more and more third party websites. Another application is the memode tool, which is a program for standardizing genome scale metabolic model into which methanetics MNX ref is playing an important role. And other application obviously are the formulation of new genome scale metabolic network. There are now a couple of applications in synthetic biology, typically in the design and engineering of new biochemical pathway. And eventually the thermodynamics of metabolism is of growing interest. At methanetics, we are also proposing a website that's give access to some model to the equation and to a range of information about the different metabolites. Here, for example, for Alpha T glucose 1 phosphate, you have data on the chemistry of the metabolite. You have the identifier and name of these compounds in other databases. And we more recently introduced a graph that allow us to navigate easily among the different exos 1 phosphate. The RDF total distribution and the sparkle endpoint of methanetics is also an access point that is more and more used. And another tool, which is built on top of MNX ref, which is MNX build that allows you to automatically build a genome scale metabolic network from an annotated bacterial genome. The development of MNX ref and the website has been made with many people, but I first like to acknowledge Sebastia Moretti and also over time many people and funding contributed to this project. Thank you for your attention.