Our goal
At twig, we firmly believe that scientists can get in the way of novel discovery. That’s because previous experiences can bias their choices. After all, it is easier to walk the well-trodden path, rather than forge a new one. That’s where the combined power of bio:drive and grow:bot can deliver big results. We set ourselves the challenge of designing and building a massive array of diverse microbes, changing pathways, enzymes, expression levels and host microbes, and demonstrating the power of full combinatorial builds.
How did we do it?
From central metabolism to palmitic acid, bio:drive recommended viable biochemical pathways for us to build. For these reactions, bio:drive selected enzymes from across a variety of natural sources, including trees, plants and yeasts. Designs were generated to investigate the expression levels of each of these pathway and enzyme combinations, resulting in more than 30 million palmitic acid microbe designs across different host microbes.
The designs were transferred over to grow:bot, for rapid building and testing. As a part of the grow:bot build process, the 30 million designs were whittled down to 4,300 for high-throughput analytics.
The result
As part of our analysis, we investigated where the top performers came from across the broad spectrum of microbe design, feeding this data directly back into our bio:drive design tool.
Above is a snippet from a combinatorial summary for one of the pathway, enzyme and expression levels we investigated – this represents only 144 of the 30 million combinations tested, with each square in the grid representing a microbe variation. Where the square is white, that combination did not make detectable levels of palmitic acid. Orange squares show ‘hits’ where palmitic acid was made, with a darker square pointing to a better yield.
The combinatorial summary demonstrates the power of diverse, full combinatorial builds to create novel, high-performing microbes. Researchers may be subject to confirmation bias as shown in the blue highlight – lots of positive signal, but not reaching the best yields possible. Whereas this simple summary shows that the best yields in the green highlight were needles in a haystack, only found through the combinatorial approach.
This data is fed back into bio:drive, improving each design, build and test cycle.