Colossal Biosciences has developed an artificial intelligence platform designed to accelerate its efforts to bring extinct species back to life, using developmental biology, comparative genomics, and three-dimensional anatomical data to identify genetic changes that could recreate traits of animals whose DNA is scarce.
The platform, called GenoMorph, is being deployed in the company’s effort to revive the dodo, an extinct bird whose limited genetic record presents a significant scientific challenge. Colossal said the technology has identified approximately 2,300 promising DNA-editing targets associated with the dodo’s craniofacial morphology, giving researchers a set of candidates to investigate and test in the laboratory.
The announcement, reported exclusively by TechCrunch, highlights how AI could help address one of the central obstacles in de-extinction research: determining which genetic differences are responsible for the distinctive physical characteristics of an extinct species when scientists have access to only a small amount of usable ancient DNA.
Register for the next Tekedia Mini-MBA.
Register for Tekedia AI in Business Masterclass.
Join Tekedia Capital Syndicate and co-invest in great global startups.
Unlike conventional comparative-genomics approaches that rely on large collections of genomes to identify genetic variations associated with particular traits, GenoMorph combines genetic information with developmental data and anatomical measurements from living birds. The aim is to narrow the search for relevant genetic changes before researchers attempt to introduce them into living organisms.
“Normally, scientists can compare many genomes to understand which genetic differences are responsible for certain traits,” Colossal CEO and co-founder Ben Lamm told TechCrunch. “We don’t have that luxury with ancient DNA from extinct species whose DNA is severely limited, and we can only get a small number of complete genomes.”
Colossal said it has obtained only one dodo genome, compared with approximately 100 genomes available for its woolly mammoth de-extinction work. The difference reflects the conditions under which the animals lived, and their remains were preserved.
Woolly mammoth DNA has survived relatively well in cold environments, while the warmer, more humid tropical conditions associated with the dodo’s habitat are less favorable for long-term DNA preservation. That leaves researchers with a much narrower genetic reference for reconstructing the extinct bird’s characteristics.
To develop GenoMorph, Colossal assembled developmental data from chickens, quail, ducks and pigeons, combining it with genetic information and three-dimensional skull measurements from more than 70 bird species. Lamm said the platform took more than a year to build and draws on the company’s existing de-extinction research.
The system analyses how genetic activity changes as birds develop, particularly in the formation of their beaks and facial structures. By comparing these developmental patterns across living species, researchers hope to identify genetic regulatory regions that may help explain the dodo’s distinctive appearance.
“The AI learns which regions of DNA are active as different birds develop their beaks and facial structures, and uses that information to predict which genetic switches may have given the dodo its iconic appearance,” Lamm said.
He described the platform as a “treasure map for the dodo genome,” explaining that it is intended to help scientists focus on genetic differences that are more likely to matter rather than investigate an enormous number of possible variations.
The approach could make research more efficient by prioritizing candidate edits for laboratory testing. However, identifying a potentially relevant genetic region is only an early stage in the process. Researchers must still establish whether a proposed change produces the intended physical characteristic, how it interacts with other genes, and whether it introduces unintended developmental consequences.
Colossal’s figure of approximately 2,300 promising editing targets therefore represents a pool of candidates for further investigation, not 2,300 confirmed genetic changes capable of recreating the dodo.
An animal’s appearance is rarely controlled by a single genetic switch. Craniofacial development can involve numerous interacting genes, regulatory elements, and developmental processes. Even if an AI system identifies regions associated with a particular trait, experimental validation remains necessary to determine which changes are causally responsible and how they behave in a living organism.
GenoMorph is intended to help the company narrow that uncertainty, but its predictions will need to be tested before they can support claims that specific dodo characteristics can be recreated reliably.
New Embryo Technology and Progress with The Nicobar Pigeon
Alongside GenoMorph, Colossal announced a system designed to deliver gene-editing tools into developing bird embryos using lipid nanoparticles. The company intends to use the technology to investigate how selected genetic changes affect development without waiting for birds to reach sexual maturity and complete an entire breeding cycle.
The technique could shorten parts of the experimental process by allowing researchers to observe developmental effects earlier. Its practical value, however, will depend on how effectively the editing tools reach the relevant cells, how consistently the edits occur, and whether the resulting developmental changes can be measured and reproduced.
The company also reported progress with its Nicobar pigeon colony, which has grown to 36 birds. The Nicobar pigeon is the dodo’s closest living relative and is central to Colossal’s plan to use a living bird as a biological starting point for introducing dodo-like characteristics.
Lamm said the company had successfully created the essential cells that develop into eggs and sperm. These reproductive cells are important because they could help researchers develop a pathway for passing selected genetic changes into subsequent generations, rather than relying exclusively on edits made to ordinary body cells.
“Because the Nicobar pigeon is the dodo’s closest living relative, those cells are a critical part of our strategy for creating birds that carry dodo traits,” he said.
The Nicobar pigeon programme, the embryo-editing system and GenoMorph address different parts of the same scientific challenge. The AI platform identifies potential genetic targets; laboratory experiments are intended to test their effects; and reproductive-cell research could eventually help establish a way to propagate selected traits.
The technologies are not yet proof that Colossal can recreate a dodo. Instead, they represent components of a research programme that must demonstrate that predicted edits can produce the desired characteristics in a viable bird.
Lamm said the company was approaching a stage at which its separate technologies could begin working together and potentially support other projects involving different branches of the evolutionary tree.
The development also raises questions about the safety and reliability of using AI to guide genetic engineering. Lamm said GenoMorph was designed to make genetic engineering “more precise, controlled, and testable,” while emphasizing that the platform does not edit DNA itself.
“Identifying a potential edit doesn’t mean we immediately introduce it into an animal or assume it’s safe,” he said. “The more precisely we can identify and validate these changes, the more informed our decisions become throughout the process.”
AI systems are increasingly moving from analyzing biological data towards influencing laboratory decisions. Predictive models can help researchers prioritize experiments, but the quality of their output depends on the underlying data, the assumptions built into the models, and the extent to which predictions hold up under experimental conditions.
For de-extinction research, the objective is also more complicated than recovering a complete original genome. Colossal’s approach involves using a living relative as a biological foundation and attempting to introduce characteristics associated with the extinct species.
The company’s latest announcement has also attracted interest from investors seeking to apply programmable biology beyond conservation. Peter Diamandis, founder of the XPRIZE competition and an investor in Colossal, announced Wednesday that he had established NeoGenesis Capital, a venture-capital firm intended to back companies using Colossal’s programmable-biology platform in healthcare, agriculture and conservation.
The planned investment activity points to a broader commercial ambition behind the research. Tools developed to identify and test genetic changes in extinct species could potentially inform other biological applications, although their usefulness in those areas will depend on separate scientific validation.