From, On the Future of Species by Adrian Woolfson
“To deride the hope of progress is the ultimate fatuity, the last word in the poverty of spirit and meanness of mind.”
– Peter Medawar, 1972
We are entering a golden age of molecular genetics, poised to form the basis of an unprecedented, biologically inspired industrial revolution in science and medicine. Biology is evolving from a descriptive science into a predictive engineering discipline that will impact almost every human endeavor. The biological sciences have, for centuries, used a reverse-engineering paradigm to decipher nature’s creations. Now, informed by artificial intelligence and large-scale genome synthesis, biology is transitioning into a forward-engineering, bottom-up, constructional enterprise. This marks a fundamental paradigmatic shift. Instead of breaking biological systems down to determine the function of individual components, we will build them to understand them.
Artificial intelligence, in combination with synthetic genomics, allows for the interrogation of vast biological datasets, the generation of hypotheses and for iteratively designing, building and testing them – linking specific sequences to defined outcomes. This process will help reveal biology’s generative rules – its hidden logic – enabling the long-sought-after mathematization of life.
In this new constructional paradigm, natural species, shaped by Darwinian evolution, will be joined by artificial species designed according to predefined human specifications. Whole-genome design and synthesis go far beyond gene editing. Whereas gene editing modifies existing genetic texts, genome writing dispenses with inherited templates altogether. It liberates life from heredity and descent, making it timeless and allowing genomes to be reimagined from scratch.
For the first time ever, species generation will no longer be dictated solely by natural selection. Rather than remaining passive observers of nature’s handiwork, we are instead becoming authors of life. The creative potential of generative biology technologies is considerable. They will transform agriculture, industry and – perhaps most significantly – medicine. We are transitioning from a Darwinian world into a post-Darwinian landscape: a limitless synthetic frontier. Like the pioneering hominins who first ventured out of Africa,
we are embarking on our own migration – into the uncharted and
borderless territory of engineered life. A landscape that we are only
just beginning to comprehend.
—–
For much of the twentieth century, the term ‘gene’ was equated with DNA sequences encoding proteins. Large stretches of DNA in the human genome with no apparent function were dismissed as ‘junk’. We now know that much of this so-called junk DNA comprises non-coding elements encoding regulatory RNA molecules, which exert a profound influence on the biology of species. These RNA-based components of biological machines are every bit as important as protein-encoding genes, and control and fine-tune their behavior.
To fully grasp the molecular logic of complex genomes, we must move beyond the protein-centric view of genomes and systematically explore their non-coding ‘dark matter’. Genome synthesis will be key to this effort, enabling the iterative rewriting and testing of specific genomic regions to uncover the rules governing regulatory sequences. We are entering an ‘RNA renaissance’, in which the role of non-coding RNA in health and disease is coming into sharp focus. This new focus on RNA will be especially critical for understanding human pathology.
——
Though biological systems differ fundamentally from digital computers, the ‘cell-as-computer’ and ‘cell-as-machine’ metaphors retain considerable practical utility. The notion of the genome as an operating system comprising a set of curated programs, is especially evident when a transplanted genome transforms the cell of one species into that of another. Yet this analogy has its limits. Many components of biological systems, for example, operate in a continuous, analogue-like manner. The logic of biology is therefore ‘fuzzy’. Furthermore, rather than being programmed, much of its order emerges spontaneously as a result of self-organizing processes. Protein folding and the assembly of viral capsids – where complex structures emerge in the absence of explicit encoded instructions – are striking examples.
Moreover, most biological components do not behave like engineered parts that perform single, well-defined functions. Instead,they often ‘moonlight’ – performing multiple functions, depending on their context. They also form the parts of complex, highly
interconnected molecular networks. To transform biology into a truly predictive engineering discipline, we will need to deconvolute this complexity. By refactoring biological systems – simplifying and modularizing their components – we may eventually render living systems as engineerable and programmable as silicon-based computing systems. The requirement for network operations may, however, set a functional limit to the extent to which this can be achieved.
——-
Biology is on the cusp of becoming fully programmable – the engineering material of choice for numerous applications. Engineering biology – variously known as synthetic genomics, generative biology, generative genomics or genome writing – is now converging with AI-augmented genome design and has the potential to address many of the world’s most pressing challenges. It promises to transform human existence and the living world in ways barely imaginable. The benefits could be immense. It could also cause incalculable harm.
The ability to design and synthesize genomes with predictive precision marks the closure of a historic biological information loop. After billions of years of evolution by natural selection, life is now at the earliest stages of decoupling from heredity. It is becoming free to reinvent itself under human direction. That we have developed techniques capable of this kind of bottom-up design may be seen as a natural corrective mechanism to address the exponential changes generated by human culture. In short, the incremental processes of biology can no longer keep pace with culture.
The velocity of cultural change driven by modern technological innovation greatly exceeds anything previously observed in human history. From the discovery of agriculture through to the invention of the printing press, internal combustion engine, computers, AI, and synthetic biology, humanity has launched itself into a state of hyper-accelerated transformation. We might call this phenomenon ‘hyperculture’ – a mode of human existence in which environmental, technological, cultural, and societal changes occur at an extraordinarily high (and exponentially increasing) frequency. The ability to design and write genomes may offer a way to reconcile biology with hyperculture – providing life with the means to update ancient biological systems designed to operate in a world that no longer exists.
——
Although still in its infancy, the science of genome design must overcome formidable technological hurdles to realize the promise of strong ABI: a fusion of Artificial Biological Intelligence capable of genuine genomic authorship, combined with the ability to construct and activate the corresponding genome sequence.
It will also require the ability to build DNA of any complexity, rapidly, efficiently, accurately, at scale, and at minimal cost. If the capacity to write genomes progresses along the same trajectory as DNA sequencing, exponential improvements are likely. Once sufficiently advanced, genome writing will enable the on-demand production of prespecified genome sequences, and even whole libraries of sequences of a particular type. Genomes themselves will be printed out routinely – like documents on a laser printer.
For now, however, we must remain imitators. Current genome synthesis efforts – even at their most innovative, for example, the creation of a novel seventeenth neochromosome by the Sc2.0 synthetic yeast project – remain principally plagiaristic. They resynthesize natural genome designs rather than inventing truly novel ones. Such engineered genomes are still constrained by the pre-existing architectures and mandates of nature. But we are moving steadily towards strong ABI, where the genuine creative authorship of unnatural genomes will become possible. The pathway to this future will be paved incrementally by increasingly capable forms of weak ABI, with progress depending on the choice of model systems, AI architectures, computing hardware and training datasets.
Bacteriophages – viruses that infect bacteria – have historically proven foundational to many of the key breakthroughs in molecular biology. They are, therefore, a logical starting point for weak ABI, providing a tractable testing ground for artificial genome design. Mastering the language of viruses will be highly consequential. It is unsurprising that the first fully artificial biological entity to be designed, physically constructed, and tested – an AI-generated Φ-X174 bacteriophage called Evo-Φ2147 – was a virus. Based on this landmark achievement, future success in the emerging field of ‘generative genomics’ will be measured not by imitating existing genomes, but by crafting entirely new ones, in order to generate biological entities that have never previously existed in nature.
REPRINT ONLY WITH PERMISSION FROM THE MIT PRESS
PLEASE CONTACT: cbirk@mit.edu
