"You Have To Iterate, You Have To Fail, You Have To Quickly Pick Yourself Up": Genome Loaded Onto Quantum Computer For First Time

The breakthrough comes from a collaboration led by the University of Oxford and the Wellcome Sanger Institute, whose researchers encoded the 1,700 base-pair genome of the hepatitis D virus into a quantum state that can be stored and processed using an IBM quantum computer and its 156-qubit Heron processor.

The aim of the project is to build quantum tools to speed up the analysis of pangenomes – datasets that capture the genetic diversity of many individuals simultaneously. Pangenomes give a more complete genetic picture than standard genomes, such as the one produced by the Human Genome Project in 2003, which drew on contributions from just a handful of individuals. 

"Having a single sequenced genome is a far cry off where we want to be," lead researcher Sergii Strelchuk at the University of Oxford told IFLScience, "precisely because we humans are very different. We need to build a data structure that captures this variation across the human population."

In 2023, the first draft human pangenome – covering 47 individuals – was published, and the Human Pangenome Reference Consortium is expected to release an even more comprehensive version covering 300-plus individuals as soon as this summer. But the richer the dataset, the harder it is to analyze.

Rather than a single linear sequence, a pangenome is better thought of as a graph or network. "It's a kind of a Tube map," said Strelchuk. The straight lines represent genomic sequences shared by everyone, and where there are loops, detours, and other alternative routes, these represent the places where individuals differ.

Mapping an individual's genome data against this structure, known as sequence alignment, involves finding which route through the map best represents their genome, and it is one of the key problems the project hopes to demonstrate could benefit from quantum computation.

This is because as the graph grows more complex, aligning a given sequence to it becomes exponentially more difficult. "You can imagine the kind of complexity of a Japanese subway-style map," said Strelchuk. 

At the extreme end, the problem becomes almost incomprehensible to classical computational algorithms: the HLA region of the human genome, for instance, which is responsible for much of our immune system's ability to fight off illnesses, produces a pangenome graph so tangled it looks like a "hairball" or a "hot mess," according to Strelchuk. "Classically, you don't even begin to hope for anything really, because it's just way too complex," the researcher said.

That's because classical computers, regardless of how clever your algorithm is, fundamentally rely on binary bits, the 1s and 0s of machine code. Quantum computers, on the other hand, are built on the concept of qubits, which differ from classical bits because of their quantum properties.

The key difference is entanglement, a property that results in each qubit not necessarily being independent of the others. They influence one another, and this means that as well as the information stored in the state of each qubit, there is also information stored in the relationships between them, called their correlations. 

As you increase the number of qubits, the number of those correlations grows exponentially, so for n qubits you have 2ⁿ correlations. Putting that into a biological context: the haploid human genome is 3.2 billion base pairs long, and as each base pair can be represented by 2 classical bits, it takes 6.4 × 10⁹ classical bits to encode the human genome. In theory, you could achieve the same with just 33 qubits.

But in practice, it’s a lot harder than just plugging a USB containing the As, Cs, Ts and Gs of a genome into a quantum computer and saying "hey presto". The genome first has to be converted into a mathematical structure that can be represented in the quantum computer, and then the team had to synthesize the precise sequence of operations that will actually prepare this state on real hardware. 

"Writing the programme, generating the circuits – this is where the biggest difficulty is," said Strelchuk. "Without generating the right instructions for a quantum computer, you really cannot achieve this goal, no matter how conceptually feasible it is."

We may have a possibility to get answers sooner, and better answers. Why not use that opportunity to do something cool?

Sergii Strelchuk

The team has been working on this problem for several years as part of an unusual, competitive research program called Quantum for Bio, which is backed by $40 million of investment from the health research non-profit Wellcome Leap. The program started in September 2023 with 12 research projects selected from a list of around 80 hopefuls. In August last year, these 12 were whittled down to the final six as part in what Strelchuk called a "Hunger Games-style" elimination, and the program concluded at the end of March, with all teams having now submitted their findings for assessment.

"The most accurate way to describe it is you run your research program as a startup," said Strelchuk. "You have to iterate, you have to fail, you have to quickly pick yourself up." He believes the pangenome team has a strong case for the $2 million prize, awarded for executing a biologically relevant task on a quantum machine of 50-plus qubits and demonstrating a clear pathway to scaling. 

A $5 million grand prize also exists, though Strelchuk thought it was unlikely that any team would meet the necessary requirements: "The depth of the program required is so big that no existing quantum computer can do it. I'll be surprised if anyone gets the $5 million prize."

While the loading of the hepatitis D virus genome is the most recent headline result, the team has achieved a range of other goals during phase three. In fact, all four of their objectives from the start of the project have been demonstrated on real quantum hardware – data encoding, sequence alignment, pangenome assembly, and phylogenetic tree construction.

 "We did all the pangenome-based sequence alignment – all four work streams have been executed on the quantum computer," said Strelchuk. "There are a bunch of papers that have been already released, and more will be released in the future."

Looking ahead, Strelchuk sees the platform the team has built as a foundation for some of the most stubborn unsolved problems in human health, including metagenomics and antimicrobial resistance. Further out lies chromothripsis – a devastating cancer mechanism in which chromosomal DNA is shattered and then reassembled incorrectly – a problem so computationally complex to understand that classical approaches have barely begun to address it.

"Classical tools are coming against this ever-increasing complexity," said Strelchuk. "We may have a possibility to get answers sooner, and better answers. Why not use that opportunity to do something cool?"