中央研究院 資訊科學研究所

Technology and resources generated by the Human Genome Project have made huge impact on research across life sciences. Our increased capacity to generate high throughput data calls for systems-level methods to decipher not only the individual impacts, but also the interactions of components in biological systems.

In the first part of this talk, I will discuss my work on modeling genetic interactions in single species, namely epistasis, which describes the non-additive effect of alleles on cellular fitness. We generated combinations of beneficial mutations that arose during the rapid adaptation of an engineered bacterium, whose growth depends upon a newly-introduced pathway. The same beneficial allele we have observed tends to become less and less advantageous when introduced to higher fitness backgrounds. We proposed a simple cost-benefit fitness model to capture how these alleles interact with each other and to quantitatively recapitulate the observed antagonistic pattern. Generalization of this mathematical model can be used to analytically predict the degree of epistasis on arbitrary fitness models, providing a new mathematical link between epistasis and pleiotropy.

In the second part, I will present a novel computational framework for studying metabolic interaction between a pair of microbes. Specifically, we computationally identify metabolites that are likely to be produced in co-culture, but that are not secreted by any of the species when growing in isolation. We found that many of the microbial communities we have tested, particularly pairs of species that are not too distant or too similar, will jointly perform such “super metabolism”.

These methods should facilitate ourunderstanding on how biological systems respond to genetic and environmental perturbations, relevant for gaining insight into evolutionary dynamics and for designing microbes for metabolic engineering applications.