What is a prediction in science?
We often think a prediction is just a guess about the outcome of an event – something we throw out into the blue and then try to test. I, for one, have definitely thrown out random guesses when I didn't do any pre-reading for my university practicals.
However, predictions are more than random guesses. They require a set of precise beliefs about how the world works. In making a prediction, we essentially pull from a past set of beliefs, particularly scientific knowledge, to say we believe this will happen in the future.
Scientific predictions must be precise so that they can be proven wrong. We like making statements that are likely to be true under all circumstances (after all, it does feel great to be right!). However, it's important to know that the point of a prediction is its provability. When our prediction is wrong, we are able to begin questioning the perceived ‘facts’ of science and revise them to more accurately predict the future. It is precisely this falsifiability, and what happens when our predictions are proven wrong or incomplete, that is sometimes more important than a correct prediction.
Essentially, we can say that scientific predictions are 1) based on past knowledge, and 2) specific.
Think of it this way: we are trying to run a printer, and we know that there is a red cartridge and a blue cartridge inside. A guess would be to say, "We will print colour”. This is not specific enough to be proven wrong, and not based on our past knowledge. Instead, a prediction would be that, based on colour theory, "if I have red and blue, I should get purple text." If we print and get green ink, I know there is something wrong with my past knowledge. What we thought was fact becomes fiction we had postulated.
Each new fact we collect reshapes the prediction that follows, nudging our proposed knowledge closer to the truth. After all, what makes science ‘science’ is the fact that it is rarely certain, but rather a process of increasingly more accurate predictions. One of the most extraordinary fields for watching this process is genomics.
Understanding our genome
The birth of genome analysis arguably came with the Human Genome Project. The idea was, in itself, quite simple. If we could sequence all three billion base pairs of human DNA, every genetic instruction our body follows, we could surely understand all diseases. It would enable us to predict, and ultimately prevent, the things that kill us.
It took thirteen years and roughly three billion dollars. It was completed in 2003 (1). It was an incredible achievement.
Unfortunately, it did not solve all problems.
However, it was incredible all the same because it revealed the scale of what we did not yet know about human function. The genome is not a simple instruction manual. It is a dense, layered system in which the same sequence can mean different things depending on context, timing, the presence of other genes, and environmental signals we are still cataloguing.
Say, if our red cartridge and blue cartridge are the genetic code, whether or not we get purple ink depends on the instructions the computer sends, whether the printer is working, and the relative amounts of red and blue ink, including the possibility of no blue at all. A green page means something is releasing yellow somewhere we didn't expect. A red page means something is blocking the blue cartridge entirely.
It's easy to know we have blue and red cartridges inside. Understanding how the cartridges are actually used by the printer has occupied genomics ever since.
These letters have allowed us to do something genuinely strange and exciting: simulate a living organism from its own genetic data. NeuroMechFly is one such project, a digital fruit fly with sensors that receive signals and six legs that respond to virtual terrain (2). The aim of the project is to find where our predictions, shown by the simulation, about real flies fail, and what that tells us about the nervous system and how it controls movement. If the model holds up to real behaviour, our understanding is roughly correct. If not, there is more to know.
So far, researchers are still working on it. But what simulations like this can do is expand our understanding of humans, since more than 60% of fruit fly genes have human counterparts, and roughly 75% of the genes known to cause human genetic diseases are also found in flies (3). Indeed, fruit flies have helped us understand human biology in the past, including giving us the first tumour suppressor gene (4).
But even so, isn't it interesting to consider the effects of using flies as a model of disease for humans? And whether our predictions about being "similar enough" actually hold? It is important to keep in mind that we cannot always extend our understanding directly across species. If we do, we can meet a paradox.
Paradoxes
We know that as we get older, we get significantly higher cancer rates. The science behind this is simple. As our cells replicate more, there are higher chances of mutations building up, thus leading to higher chances of cancer. Similarly, larger humans have higher rates of cancer, simply because there are more cells, more divisions, and as a result, a higher chance of mutation.
Essentially, printing a thousand pages should produce more errors than printing ten.
If we extend this knowledge to a bowhead whale, which weighs over 80,000 kg and lives more than 200 years compared to humans – averaging around 70 kg and 80 years ourselves – we would predict whales must be riddled with cancer.
This prediction was reasonable, evidence-based, and most importantly, wrong.
Across mammals, there is no correlation between body size, lifespan, and cancer incidence. A mouse weighs around 20 grams. A blue whale weighs up to 150,000 kilograms. The cancer-rate difference between them is incredibly small.
The observation that large, long-lived animals do not carry proportionally enormous cancer burdens was first formally noted by the statistician Richard Peto in 1977. It became known as Peto's Paradox, and it has been generating arguments ever since (5).
I have two predictions about why it fails. First, cancer rates in large animals are simply difficult to measure. Our data is incomplete, our sample sizes are skewed, and our conclusions are premature. Second, for a whale-sized organism to exist at all, it must have evolved mechanisms of cancer suppression far beyond our own. Either or both could be true. Neither might be.
So far in science, we have found extraordinary mechanisms that potentially explain parts of this paradox.
TP53 is a tumour suppressor gene which codes for a protein that detects DNA damage and triggers cell death before a damaged cell can divide. Elephants have 20 copies of this gene compared to the single copy humans carry (6). When their cells are exposed to DNA damage, they undergo apoptosis more effectively, meaning problematic cells die before they can replicate.
Bowhead whales express a gene called CIRBP at around 100 times the level of other mammals, which dramatically improves the repair of double-stranded DNA breaks (7). Rather than killing damaged cells, the whale's cells repair the DNA so that mutations rarely accumulate.
When researchers applied these mechanisms by introducing higher CIRBP expression into fruit flies, the flies lived longer and became more resistant to radiation-induced DNA damage. In human cells, DNA repair efficiency roughly doubled (7).
We are starting to take what evolution spent millions of years building and turn our predictions into intention. We are starting to explore the possibility of applying these mechanisms to humans through the exciting field of genetic engineering.
Genetic Engineering
If we understand genetic information from genomics well enough, can we rewrite it?
As with all things in science, the answer is: only if we narrow the scope of the question, and provide specificity.
If a genetic condition is caused by a single gene, the answer increasingly is yes.
One such case is a baby called KJ Muldoon, who was born with a faulty gene for the protein that breaks down nitrogen. Without it, ammonia builds up in the blood, which is highly toxic. Using a CRISPR base editing therapy designed specifically for KJ's mutation, the wrong DNA letter was converted to the correct one (8). Instead of living in and out of hospital, KJ has been able to live a normal life.
Genetic engineering is a field of extraordinarily rapid development. We have already moved from the early approach of extracting cells, editing them outside the body and returning them, to delivering the editing machinery directly into living tissue. Maybe one day we can answer the broader question of addressing diseases that are caused by far more complex factors than a single gene. But for now, being able to help KJ live a normal life despite being born with a condition that had no cure is an inspiring example of how predictions can be turned into intentions, fiction into fact, and how genomics is changing the world of biology.
References
National Human Genome Research Institute. International consortium completes Human Genome Project. National Institutes of Health. April 14, 2003. https://www.genome.gov/11006929/2003-release-international-consortium-completes-hgp
Lobato-Rios V, Ramalingasetty ST, Özdil PG, Arreguit J, Ijspeert AJ, Ramdya P. NeuroMechFly, a neuromechanical model of adult Drosophila melanogaster. Nat Methods. 2022;19(5):620-627. doi:10.1038/s41592-022-01466-7
Reiter LT, Potocki L, Chien S, Gribskov M, Bier E. A systematic analysis of human disease-associated gene sequences in Drosophila melanogaster. Genome Res. 2001;11(6):1114-1125. doi:10.1101/gr.169101
Gateff E. Malignant neoplasms of genetic origin in Drosophila melanogaster. Science. 1978;200(4349):1448-1459. doi:10.1126/science.96525
Callier V. Core Concept: Solving Peto's Paradox to better understand cancer. Proc Natl Acad Sci U S A. 2019;116(6):1825-1828. doi:10.1073/pnas.1821517116
Sulak M, Fong R, Mika K, et al. TP53 copy number expansion is associated with the evolution of increased body size and an enhanced DNA damage response in elephants. eLife. 2016;5:e11994. doi:10.7554/eLife.11994
Firsanov D, Zacher M, Tian X, et al. Evidence for improved DNA repair in the long-lived bowhead whale. Nature. 2025;648(8094):717-725. doi:10.1038/s41586-025-09694-5
Children's Hospital of Philadelphia. World's first patient treated with personalized CRISPR gene editing therapy at Children's Hospital of Philadelphia. CHOP News. May 15, 2025. https://www.chop.edu/news/worlds-first-patient-treated-personalized-crispr-gene-editing-therapy-childrens-hospital
