The demise of Ginkgo Bioworks has had far reaching consequences beyond those suffered by their shareholders. The share price is now 1-2% of the peak value but it has coincided with a huge withdrawal of investment capital from biotechnology in the wake of the pandemic. Yet the cost of healthcare in the West moves ever closer to unaffordability and the potential of biomanufacturing to drive huge economic growth using more sustainable methods remains.
In a 2021 report Scorpion Capital alleged Ginkgo Bioworks Holdings Inc to be a “colossal scam”, claiming that the company, at a $23 billion valuation, was ridiculously overpriced. Former employees cited in this short report claimed that 9 out of 10 projects launched by the company since its inception have failed.
We maintain that rather than being a “colossal scam” the demise of Ginkgo was a predictable and inevitable consequence of the realities of evolution and what the genomics of evolution teaches us. If the lessons from nature had been assimilated into the Gingko approach the hype could have been converted into delivery.
Whether the inevitable scepticism in the wake of Gingko’s high profile demise has dealt a long-term or fatal blow to the access to capital for the synthetic biology movement remains to be seen. However, finally delivering results to restore faith need not be a bank breaking exercise.
QTL technology, being pioneered by Phenotypeca, offers just such redemption.
Biofoundry model doesn’t deliver
Ginkgo describes itself as a biofoundry, where key foundational data and capabilities can be used for aspects of multiple R&D projects without the need for repetition. The commercial model includes fees and a share of the upside from products that Ginkgo is involved in creating. Ginkgo also uses synthetic biology to create strains for bioprocessing manufacture and biosecurity.
The company’s hypothesis, as set out in this video, is that by taking a modular engineering approach to biological research similar to that deployed in computing, development can be accelerated and optimised to obtain a much higher probability of success. The video shows a pyramid of DNA base pairs at the bottom, leading to successive layers up the pyramid of genes, proteins, pathways and ultimately phenotype. This is then compared to a similar pyramid that shows how computing and computer engineering have evolved.
Breakthrough medicines haven’t arrived
The problem is that AI and synthetic biology, based on finding correlations and patterns in large data sets, do not establish mechanisms for breakthrough medicines. They also don’t offer a means of developing commercially competitive and scalable strains for biomanufacturing of established products. This bar is even higher when such products are already in mature and competitively priced markets.
Why? Because biology is not modular. This fact has profound implications for the entire synthetic biology movement, whose ambitions extend to delivering fundamental efficacy breakthroughs in highly complex diseases such as degenerative and metabolic conditions.
The genome is not modular
Biology is not engineering, nor is it software. Its structure is the result of millions of years of evolution as species adapt over many generations to their environments. Survival of the fittest has eliminated the designs that are particularly prone to damage commonly arising from daily interactions. This means human interactomes have compensatory pathways. In other words, they are wired so that critically important functions can be maintained if environmental damage compromises the primary pathway. This has profound implications for the efficacy – or otherwise – of drug development.
The genome is dizzyingly complex
The human cell has over 3 billion bases pairs in its genome and, on average, from their closest relatives the genome of such cells varies around every 120 bases.. A phenotype can refer to different cells in the same individual, different individuals or a disease cell in comparison to its healthy equivalent. The combination of variations in genomic bases, and their interactions that result in the phenotype of interest, are not in predictable pathways that can be easily identified by standard heuristic research methods.
For context, the number of atoms in the universe is around 1081 whereas the number of experiments to establish the optimal pattern of variants in the genome, for a desired functional phenotype of interest, would be around 1030,000. This is far more than any future AI could handle, without even taking into account the experimental work needed to verify their validity. The only way past this complexity is to use the tools of evolution itself.
Harnessing the power of evolution
The fuel of evolution is relevant genetic diversity. Thereafter, the patterns of variation undertaken by evolution cannot be predicted. However, if sufficient progeny are produced, then phenotypes that more closely resemble the phenotype of interest can be generated. Relevant screens can identify these better adapted phenotypes, which can then be subjected to specialist QTL analysis, determining which genomic variants are responsible for their more optimal characteristics.
Iterative breeding beats sequential engineering
Thereafter the optimum strains can be selected as parents, and the process of breeding, screening and specialist genomic analysis repeated. Iterative cycles continue until the optimum phenotype is obtained. Within this approach evolution is making variant combinations simultaneously, with the selection/screen for optimum strains providing successful combinations in a few rounds of breeding. This contrasts with sequential engineering of individual elements, which takes infinite time, even with the high throughput individual changes that are possible with synthetic biology and engineering.
Alternatively, the relevant phenotypes can be the starting point, with QTL analysis used to identify the genomic traits responsible for its specific functionality. This information then informs specialist complex interactome analysis to identify the locations of high importance for the biological function of interest for the drug discovery program in question.
Transformative promise
The successful deployment of QTL technology for strain optimisation has already been used to create optimum manufacturing strains for defined products. Proprietary complex interactome technology has also been used to analyse biological cells and systems, delivering efficacy breakthroughs for drug and other product discovery. This contrasts sharply with the AI and high-throughput engineering approach borrowed from software and engineering as used in the mainstream approach of synthetic biology. An approach with an extremely low prospect of success for the reasons outlined above.
Ginkgo was arguably the flagship firm for synthetic biology but its fall from grace has equally dragged the perception with it. Billions of years of evolution has demonstrated that biology is not modular and cannot be unlocked by engineering techniques no matter how powerful the AI tool driving them. Redemption is available and, perhaps ironically, in the form of nature itself and how it chooses between those that prosper and those who are consigned to history.
