Browsing by Subject "Transcriptome analysis"

Now showing 1 - 1 of 1
  • Thumbnail Image
    Item
    Systems analysis of complex biological data for bioprocess enhancement.
    (2008-12) Charaniya, Salim Pyarali
    Recent advances in data-driven knowledge discovery approaches, such as `omics' technologies, provide enormous opportunities to uncover the multifarious determinants of several pharmaceutically relevant biological traits. This work focuses on the challenges, which include: (i) Deciphering the regulation of antibiotic production in Streptomyces coelicolor, and (ii) Elucidating the attributes of high recombinant protein productivity in mammalian cell culture processes. The phenotypic complexity of Streptomycetes, which produce several clinically relevant antibiotics and other natural products, manifests in their diversity of secondary metabolism and morphological differentiation. To identify the dynamic gene regulatory networks that confer such complex phenotypes, the temporal transcriptomic characteristics of the model organism S. coelicolor, under more than twenty-five diverse genetic and environmental perturbations, were integrated with other functional and genomic features. A whole-genome operon map was also predicted, and a significant portion of the map was experimentally verified. Such a systems approach can reveal several insights about the functional processes relevant for antibiotics production. The therapeutic value of recombinant proteins has brought about a continuously rising demand that is met by development of hyper-producing mammalian cell lines. However, the molecular ingredients of high productivity are not well understood. The transcriptomes of several recombinant antibody-producing NS0 cell lines with a wide productivity range were surveyed in an attempt to identify the physiological functions that are modulated in high-producing cells. Cell culture process enhancement also entails an understanding of the process parameters and their interactions, which are critical determinants of high recombinant protein productivity. The comprehensive process archives of modern production plants present vast, underutilized resources containing information that, if unearthed, can enhance process robustness. The on-line and off-line process data of several production `trains' from a commercial manufacturing facility were investigated using kernel-based machine learning tools to elucidate predictive correlations between process parameters and the outcome. Together, such discovery strategies based on integrative data mining hold immense potential for enhancing our understanding of industrially relevant biological processes.