Genetic drivers of protein changes over time: Findings, considerations, and approaches in TOPMed cohorts and UK Biobank
This study utilizes longitudinal proteomics data from TOPMed and UK Biobank cohorts to demonstrate that while baseline-adjusted models identify numerous genetic drivers of protein changes over time, many of these findings likely stem from statistical artifacts like collider bias or regression to the mean, highlighting the critical need for careful modeling strategies to accurately uncover true genetic mechanisms of proteomic aging.
Original paper licensed under CC BY 4.0 (https://creativecommons.org/licenses/by/4.0/). This is an AI-generated explanation of a preprint that has not been peer-reviewed. It is not medical advice. Do not make health decisions based on this content. Read full disclaimer
Imagine your body as a bustling city. For a long time, scientists have taken snapshots of this city to see what's happening right now. They've mapped the population (your genes), counted the buildings (your proteins), and noted the traffic jams (diseases). But a city is never static; it's always changing. Some buildings get taller, some roads get wider, and others crumble. The big mystery in modern science is: why do some cities age gracefully while others fall into disrepair? We know that time passes for everyone, but the rate at which our biological city changes varies wildly from person to person.
To understand this, scientists look at "proteins." Think of these as the workers, construction crews, and messengers running around your body's city. They do the heavy lifting, fix the pipes, and send signals. We also know that our DNA (the city's master blueprint) influences how these workers behave. Usually, scientists just check the blueprint to see how many workers are present at a single moment in time. But what if the blueprint also dictates how fast the number of workers changes over the years? That's the question this study tackles: Can we find the genetic instructions that control the speed of change in our body's workforce, rather than just the size of the workforce at one specific moment?
The Great Protein Race: Who's Speeding Up and Who's Slowing Down?
In this study, researchers decided to play a game of "spot the difference" over a very long time. They gathered data from three different groups of people (cohorts) who were checked on their protein levels at multiple visits over roughly 12 to 18 years. It's like taking a photo of a runner every few years to see if they are speeding up, slowing down, or staying the same pace.
The team looked at nearly 3,000 different proteins—the body's workers—and tried to calculate a "slope" for each one. A slope is just a fancy math word for a rate of change. If a protein level goes up steadily over time, it has a positive slope; if it drops, it has a negative slope. The big question was: Do our genes control these slopes?
The "Before" Trap
Here is where things got tricky, and the researchers found a major pitfall. When they first tried to find the genetic drivers of these slopes, they ran into a problem with how they did the math.
Imagine you are trying to figure out if a specific type of shoe makes people run faster. If you only look at the people who already have fast shoes, you might get confused. Similarly, the researchers found that if you try to predict how fast a protein will change while also knowing exactly how high that protein started, the math gets messy. It's like trying to guess how much a plant will grow in a month, but you're also forcing the math to account for how tall the plant was on day one.
When they did the math without looking at the starting height (baseline), they found very few genetic clues—only 19 proteins showed a clear genetic link to their rate of change. But when they did look at the starting height, the number exploded to over 600 proteins!
The "Magic" Illusion
The researchers suspected this explosion of 600 proteins might be a trick of the light, a statistical illusion known as "regression to the mean" or "collider bias."
Think of it like a classroom test. If you pick the students who scored the lowest on the first test and tell them, "We are going to see how much you improve," you might find that their scores go up. But that doesn't mean a specific gene made them improve; it just means they started low and naturally moved toward the average. The researchers realized that by adjusting for the starting level, they might have accidentally created a fake signal.
To test this, they tried to replicate their findings in two other groups of people (the UK Biobank and CARDIA).
- The 19 proteins found without adjusting for the start line? They barely showed up in the other groups.
- The 600 proteins found with the adjustment? They showed up a lot, but the researchers are very cautious. They found that many of these signals seemed to be the result of the math trick rather than a real biological driver. In fact, they found that the "adjusted" results often pointed in the opposite direction of what the "unadjusted" results suggested, which is a huge red flag in science.
The Verdict: It's Complicated
So, what did they actually find?
- The "Real" Drivers are Rare: The study suggests that finding the genes that truly control how fast proteins change over time is incredibly hard. They only found a handful of strong candidates (about 19) that didn't rely on the tricky math adjustments, but even those were hard to confirm in other groups.
- The "Fake" Drivers are Everywhere: The hundreds of genetic links they found when adjusting for the starting level might be mostly statistical noise. The paper argues that this method might be creating false alarms, making it look like genes control change when they might just be reacting to where the protein started.
- A Surprising Connection: While the direct "slope" signals were hard to pin down, the researchers found a strong link between how proteins change and how much they vary. They discovered that 73% of the proteins that showed genetic control over their variability (variance) also showed genetic control over their rate of change (slope). This suggests that the genes influencing how much a protein fluctuates might be the same ones influencing how fast it trends up or down, even if the specific "slope" math is tricky to interpret.
Why Should You Care?
The researchers aren't saying they solved the mystery of aging. Instead, they are sounding a cautionary bell. They are telling the scientific community: "Be careful how you measure change!" If we use the wrong math, we might think we've found the genetic keys to slowing down aging when we've actually just found a mathematical illusion.
The study highlights that while our genes definitely influence our bodies, pinning down exactly which genes control the speed of change is a much harder puzzle than just finding genes that control the amount of protein at one moment. It's a reminder that in the race to understand aging, the finish line is tricky, and sometimes the fastest-looking runners are just running in circles.
Drowning in papers in your field?
Get daily digests of the most novel papers matching your research keywords — with technical summaries, in your language.