Metabolic network reconstruction: tomato fruits as a model.

KAMENETZKY, L.; LÓPEZ, M.; STEGMAYER, G.; MILONE, D.; SORREQUIETA, A.; VALLE, E.; GIOVANNONI J.; FERNIE A. R.; CARRARI, F.

Workshop; EU-SOL meeting; 2010

The ‘*omics’ technologies make essential the integration and analysis of heterogeneous information sources in order to describe and reconstruct complex biological networks. Because the integration of such large amounts of data from different origins requires the automation of statistic calculations and visualization tools, we developed a software  named *omesom (Milone et al., 2010) based on self-organizing neural network model (Stegmayer et al., 2009). The aim of this work is to facilitate the discovery of new relationships between genes and metabolites using the developed software. Starting with tomato introgressed lines (ILs) from S. lycopresicum x S. pennellii as biological model, transcriptomic and metabolic data were obtained. After preprocessing, filtering, selection and normalizing steps, 1159 transcripts and 70 metabolites from IL tomato cultivars were linked to currently available databases (SGN http://solgenomics.net/ and KEGG, http://www.genome.ad.jp/kegg/, for transcripts and metabolites respectively). The data were integrated and analyzed with *omesom software that provides an easy identification of data clusters that group together both metabolites and transcripts having similar measured values in the ILs of interest.  An interesting feature found in this work is transcripts and metabolites from amino acid metabolism (GABA, Glutamate, putative GABA transporter, among others) clustered together in neighboring neurons. In particular, a region in chromosome 2 already described with genomic variability between both Solanum species (Kamenetzky et al., 2010) was corroborated to have QTLs from this specific amino acid metabolic pathway by enzymatic activity. We demonstrated that neural network analysis is a valuable strategy for reconstruction of metabolic pathways