AI for Synthetic Biology: IWBDA’18

Living as we do in the age of artificial intelligence it came as no small surprise that AI formed a common thread throughout much of the International Workshop on Bio-Design Automation, which this year celebrated its 10th anniversary in Berkeley, CA. However, while the potential for AI technologies grows daily, applications of AI within synthetic biology have lagged behind.

Héctor García Martín’s insightful keynote placed the blame for this deficit firmly on the difficulty of generating good quality data in sufficient quantities for machine learning. In his own work, he uses AI to improve predictions for increasing yields in biofuel production, and relies increasingly on automation to increase data generation and to capture valuable metadata associated with each run.

This year the IWBDA opened with a workshop demonstrating the latest features of the Synthetic Biology Open Language (SBOL) standard, which saw a great increase in functionality last year with the release of version 2.2. SBOL is a file format for the modular design and description of genetic systems, which is primarily designed as an export and interchange format for BioCAD tools. It is thus, highly complementary to the capabilities of our own open source language for biology, Synthace, which integrates inputs from a variety of in-silico BioCAD packages, before enabling the in-vitro execution of them physically in the user’s own laboratory. Nicholas Roehner presented a poster demonstrating how SBOL can be leveraged to represent combinatorial designs, with new support for templating, while Bryan Bartley presented a case study of how the Standard European Vector Architecture (SEVA) database was converted into an SBOL compliant database using SynBioHub. Bryan also presented the vision for the future of SBOL, demonstrating the new features of version 2.2 which supports tracking of the engineering lifecycle through generic Activities.

Attendees I spoke to seemed largely divided on the progress of SBOL, with some seeing the development of a common format as highly important if not essential for the maturation of the field as a whole, while others feared it would act as a technical barrier to publication. Key contributor and member of the SBOL steering committee Jake Beal addressed these concerns straight on, highlighting that SBOL and its associated libraries are not intended to be used directly by scientists, but as a common file format for platforms and tools to communicate.

It seemed to me that there is some considerable merit in this view. The ecosystem of tools for synthetic biology continues to grow, driven by the desire for high quality, well-annotated data, users are likely to expect to move their data seamlessly from tool to tool. While scientists are often put off SBOL due to its perceived complexity, it may yet help bridge the gap between different platforms.