Two computational simplex approaches to graphical highlighting metabolic phenotypes and their functional origins: correspondence analysis and weighted metabolic profiles analysis
Résumé
The metabolism is a complex system interacting with several intrinsic and extrinsic factors of biological organisms, viz. genome expressions, physiological states, environmental conditions, etc. This multifactorial interaction means that the metabolism works as a reactive and flexible system providing reliable biochemical pictures on the effects of different governing factors. Metabolic flexibility and reliability are linked to conservation laws, constraining the metabolism to a close system linking input (resources) to output (products) signals: any entering signal will be decomposed into weighted parts through different metabolic pathways. This gives to metabolic trends different functional degrees highlighted by different relative levels of metabolites. Sharing the same unit resource, the different metabolic pathways are statistically constrained to be regulated within a simplex space characterized by a unit sum of its components. Output metabolic responses and their inside regulatory processes can be analyzed by using two simplex-based approaches: correspondence analysis (CA) and weighted metabolic profiles analysis (WMPA), respectively. These two approaches are based on two opposite (complementary) principles consisting of decomposition and combination of metabolic variability. In CA, metabolic datasets are decomposed into extreme trends representing elementary components of metabolic polymorphism called metabotypes. In WMPA, iterated combinations between different metabolic components help to extract functional information on their generator backbone system.