Industrial engineering is difficult enough. Now imagine the challenges of bio-engineering, where your factories are living cells, with all of the thousands upon millions of interacting factors involved. How do you keep track of everything? How can you determine whether the changes you want to make will improve yield, or have drastic consequences somewhere else down the line? How can you follow up on the mountains of data which are inevitably produced during this process?
Providing the tools to answer these questions is Synthace Ltd. of London.
Synthace (which will be at SynBioBeta SF 2015) has recently been moving from strength to strength, recently announcing a collaboration covering automated microbial strain engineering with the agricultural giant Dow AgroSciences LLC. This collaboration includes a multi-seat enterprise license allowing the company to use Synthace’s Antha software tools, which will likely form the backbone of a program to develop improved microbial fermentation strains for the production of crop protection products.
Dow AgroSciences, a subsidiary of the Dow Chemical Company and with global sales of USD 7.3 billion in 2014, is heavily involved in the production of compounds to boost crop yields – pesticides, herbicides, fertilisers, etc. A number of these compounds are now being made by engineered microbial strains – thus maximising yield requires the balancing of multiple factors to develop highly efficient bio-producer strains. When multiple factors are in play, single experiments rapidly become too time-consuming. Worse than this, the difficulty of biological science inevitably means that non-standard approaches begin to make their appearance.
Research and development thus requires a standardised, flexible, multi-factorial system. This is where the collaboration with Synthace, and their Multifactorial Experiments for Systematic Augmentation (MESA) approach, come in. MESA is considered a ‘high-dimensional’ approach, based around performing hundreds of parallel experiments with varying factors. Statistical analysis allows the important factors controlling yield to be identified for further optimisation, while the less-important can be put to one side.
Obviously all of these parallel experiments generate vast amounts of data, and thus Synthace has heavily invested into the field of laboratory automation. Their software and tools are designed to interface with automated systems, from lab robots through to industrial-sized fermenters, for both control and data analysis. The key to this flexibility lies in Antha, a biological programming language.
Source: Synthace https://www.antha-lang.org/docs/intro.html
Antha is designed to act as an open-source programming language for biological experimentation, and one which can easily link into automated laboratory systems. Sounds impressive, but what does it actually mean? Imagine a computer program. A computer program is essentially a list of instructions, a way of taking input data (say, the current temperature, in Fahrenheit), performing an operation of calculation with that data (subtract 32, multiply by 5, then divide by 9), and then output the result of that operation (the current temperature in Celsius). A modern program will use thousands upon thousands of these variable inputs, which will often be bundled together into related groups – so Weather would have Weather.Temperature, Weather.Humidity, Weather.Sunny, etc.
How does this apply to bioscience? Think of a typical 50ml tube. It has a number of attributes, some useful (e.g. Tube.Contents is humanised antibody solution, and Tube.CurrentVolume is 45ml) and some less so (Tube.CapColour is blue). If we are going to perform an experiment with the antibodies within this tube, we need to take the input data (Tube.Concentration of 20mM), perform a calculation (what is my desired final concentration?) and action (dilute the solution), then return the result of that process (the diluted solution). A vast amount of experimentation comes down to input, operation, output, and thus matches exceptionally well to programming languages.
More than that, Antha is a multi-level programming language, which means that the nitty-gritty details can be ignored in favour of a broad overview. Take the example above, in which we want to dilute our solution. This is an action we’ve done many times, in many circumstances, and can consider as an extremely generic procedure, along the lines of Dilute[Dilute what?][From what?][To what?] – more specifically Dilute[The blue tube][From 20mM][To 3µM]. How does it get diluted? Lab robot, long-suffering PhD student? At this level, it doesn’t really matter, when designing our experiment we simply need to know that it has been done. This means that researchers can sit back and design the overall flow of the experiment, with the details of the process (programming the lab robot, buying the student a beer) can be specified at a lower level.
This, in essence, is programming. The reason for Synthace’s success is that their Antha language and associated tools allow for development to be standardised (high-level) and adapted to individual laboratory or production conditions (low-level). This flexibility is highly prized in the cut-throat world of industrial bio-engineering, allowing the development of hardware-agnostic protocols for numerous roles. As the recent agreement with Dow AgroSciences for development of automated strain development clearly shows, this system can cover extremely complex circumstances.0