Two Meters of DNA in a Soccer Ball: A Harvard Structural Biologist on the Machines That Read Your Genome

Introduction: A Conversation About a Field in Transition

Structural biology is the discipline where questions about how life works get answered at the level of shape: what a molecule looks like, how its parts move, and how that geometry determines what it does inside a living cell. It is also a field being reshaped by computation.

Lucas Farnung, a structural biologist and assistant professor of cell biology in the Blavatnik Institute at Harvard Medical School, spoke with Harvard Medicine News about how machine learning is accelerating research in his area. The conversation ranged across the machinery of transcription, the sheer physical problem of packing a genome into a nucleus, and the unglamorous but essential question of how a lab actually obtains the molecules it wants to look at in the first place.

What Farnung's Lab Studies: The Machinery of Transcription

Farnung's lab investigates the structure and function of the molecular machines responsible for transcription — the process that regulates gene expression.

Transcription sits at the front of a chain of events that keeps an organism running. During transcription, molecular machines read the instructions contained in DNA and create RNA. Other molecular machines then read that RNA and use the information it carries to make proteins, which fuel almost all activities in the body.

Within that broader process, the lab has a particular target. Farnung's group is especially interested in visualizing how the molecular machine RNA polymerase II gains access to genomic information and transcribes DNA into RNA.

The Packing Problem: Two Meters of DNA in a Few Microns

Part of what makes RNA polymerase II's job difficult is a problem of pure physical scale. Stretched out end to end, the human genome would run roughly two meters — about six and a half feet — in length. All of it has to fit inside a cell nucleus only a few microns across.

Farnung offers an analogy for the mismatch: it is like squeezing a 150-mile fishing line into a soccer ball.

Cells manage the feat by compacting DNA into a structure called chromatin. But that solution creates a conflict of its own. Once DNA has been packed away into chromatin, the molecular machines that need to read it can no longer easily access the genomic information stored on it. The cell has solved a storage problem and, in doing so, created an access problem — one that the machinery of transcription has to overcome every time a gene is expressed.

How the Molecules Get to the Microscope

Before any of these machines can be visualized, they have to be obtained in usable quantities — and that is a laboratory project in itself.

The lab's experimental approach involves isolating molecular machines from cells and then examining them with specific types of microscopes or X-ray beams. Getting to that point works roughly like this:

• Genetic material coding for the human molecular machine of interest is introduced into insect or bacterial cells.
• Those cells produce large amounts of the machine.
• The machine is then purified out of that material.

This production-and-purification pipeline is the gateway step. No structure gets visualized until it has been completed.

The Scale Bottleneck That Motivates New Tools

The gateway step is manageable for one molecule. The trouble is that transcription is not regulated by one molecule.

Thousands of proteins interact with each other to regulate transcription. Understanding those protein-protein interactions experimentally means running the purification process not once but thousands of times — the same sequence of introducing genetic material, growing cells, harvesting product, and purifying it, repeated over and over.

That arithmetic is the clearest illustration of why computational acceleration holds such appeal for structural biologists. When the experimental path to a complete picture involves thousands of iterations of a labor-intensive protocol, any tool that can shorten or short-circuit parts of that loop changes what a lab can realistically attempt.

A Note on Sourcing

Reporting for this piece drew on the Harvard Medicine News interview with Lucas Farnung.

Three additional PubMed Central pages supplied for this story returned only a reCAPTCHA browser-check page and yielded no retrievable content on the subject. No claims in this article rest on them.

Readers should therefore treat the scope of this article as limited to what the Harvard Medicine News conversation established.