Section: New Results

Bespoke tools for comparative genomics

Large-scale comparison of strains of cell factory species is an indispensible tool for understanding the genetic origin of phenotypic variability, and can considerably optimize the selection and construction of high-performing industrial strains. For example, in oenological applications new strains may be selected based on their influence on aroma, their adaptation to grape musts, or their robustness during fermentation. In oil production applications, new strains may be selected based on their yield, or on the saturation degree of the lipids, or on their growth characteristics. Comparative genomics has proven quite effective in understanding cell factory diversity [1], [6], [5], [36], [31]. A typical project will involve 500 segregants and 50 genomes. Accurate and rapid analysis of the concomitant data volumes requires efficient tool sets that must be adapted to the real use cases of the industrial application.

Pleiade addresses this problem through the definition of bespoke software systems that associate integrated sets of tools, including its Magus software platform (section 6.1). A key challenge in defining this kind of integrated system is the need to connect the components. We develop configuration formalisms whose solutions are orchestrations of weakly-coupled microservices running in independant containers. These services may be data banks, genome browser and visualization software, workflow tools like Galaxy, machine learning algorithms for classification, or shared workbooks like Jupyter or Zeppelin. By formalizing the connections between services, we can simplify deployment, and also create an opportunity for continuous deployment.