María Andrea Angarita Rodríguez ¹ Nicolás Mendoza-Mejía ¹ Jason Papin ² Andres Felipe Aristizabal-Pachon ¹ Andres Mauricio Pinzón ³ Janneth Gonzalez ^1*

• ¹ Pontifical Javeriana University, Bogotá, Colombia
• ² University of Virginia, Charlottesville, Virginia, United States
• ³ National University of Colombia, Bogotá, Bogota, Colombia

The increase and availability of large-scale multi-omic data have allowed us to study cellular machinery systematically. However, few studies take advantage of the potential of the biological knowledge extractable from these data and their full integration into Genome-Scale Models of Metabolism (GEMs). The most common practice is associating transcriptome and proteome data independently to reaction boundaries, providing the model with an approximation of the maximum reaction rates based on each omics data. However, the use of these sources separately usually incorporates uncertainties and inaccuracies into the model. Thus, to overcome some current limitations we have applied a recently developed, principal component analysis (PCA)-based approach to integrate transcriptome and proteome data, with the aim to reconstruct context-specific models based on multi-omics data. Moreover, following this approach we also reconstructed an astrocyte GEM that has improved prediction capability over state-of-the-art astrocyte GEMs available in the literature. Such models will play an essential role in the study of neurodegeneration and the development of effective therapies.