Applying Dynamical Systems Theory to Improve Analysis of Time Series Transcriptional Data from Bioproduction Processes

Scaling bioproduction processes from laboratory and pilot scale systems up to commercial scale is one of the principal problems facing the new bioeconomy, and one towards which the Department of Energy and other government agencies have invested substantial resources. To address this problem, this SBIR project proposes research to develop process control systems to ensure repeatability, efficiency, and predictability of bioproduction systems. In Phase I of this SBIR project, Mimetics, LLC (“Mimetics’) compiled and curated a library of whole transcriptome RNA-seq time series gene expression data from biofuels fermentations driven by Saccharomyces cerevisae (‘yeast’). We also designed, coded, and tested algorithms to compare time series gene expression profiles and samples across fermentations, interrogating this large database. Machine learning algorithms were then deployed on the results of these dynamic gene expression comparisons. The algorithms enabled us to use data from a sample from a single time point within a fermentation process to identify states and phases for the entire fermentation. The research proposed in this Phase II project will focus on extending the database and algorithms developed in Phase I, further testing them for reliability, and embedding them into fully automated pipelines. Extending the database of fermentation data and including data from a variety of fermentation conditions will expand the statistical power and diagnostic ability of our algorithms. Existing algorithms will also be adapted from ethanol-producing fermentations to isobutanol-producing fermentations for bio-jet fuel applications. These algorithms and pipelines will be wrapped in a web-based graphical user interface for uploading, analysis, and visualization of data and results. The result of this Phase II project will be a complete suite of fermentation analysis tools, easily accessible to both R&D and production personnel in companies and laboratories developing and producing biofuels for many applications. The algorithms will be the basis of a Software as a Service’ system to be made available on a one-off or a subscription basis to industry, government, and academia to enable much more rapid design and scale-up of bioproduction systems. As the business expands, the database will be expanded from Saccharomyces to other fungi and bacteria used in bioproduction processes, and the algorithms will be adapted for these applications. This SBIR project will be the basis of a company offering a suite of software tools to the bioproduction community to enable more rapid progress from the lab to the market.