Outliers Matter: Bradford Shifts Data From Statistical ‘Noise’ to Biological Mechanism

Kent Bradford is offering a new way of looking at biological data that, he says, “radically alters what we can see.” 

Traditionally, scientists see things in terms of averages to understand the behavior of plants, cells, molecules – all kinds of things. But now, the UC Davis distinguished professor emeritus argues, new technologies allow us to see data points as individuals and take a closer look at the statistical outliers. Usually dismissed as “noise” and “error,” they in fact offer a goldmine of new insights about how things respond to a stimulus.

That’s because, Bradford proposes, individual things in a population respond to any given input along a range of trigger points, and life has harnessed that variation to its advantage.

Bradford calls this new viewpoint the population-based threshold model. At its heart, it posits that individuals in a group are not all the same, even at the level of cells and molecules.

“This could very well be the way everything in biology works,” said Bradford, based in the Department of Plant Sciences. “We just have to put on our population-colored glasses to see it.”

In his new book, just released by Oxford University Press, Bradford shows how a simple formula he developed – originally based on his observation of seeds – can explain the behavior of things as diverse as plant enzymes, human pancreas cells, even people at the broad level of economics and politics.

The model is made possible by new technologies, developed just in the last decade or so, that allow scientists to see individual things with unprecedented detail. 

In addition, Bradford’s formula reveals how statistical outliers form subpopulations, altering the shape of the traditional bell curve. Bradford further encourages fellow scientists to test his hypothesis with their own data.

His book is, “Populations of Individuals: Understanding Biological Variation from Molecules to Ecosystems.” You can find it here.

Variation: A basic feature of the system

Bradford began to ponder this idea in the 1980s as an early career scientist at UC Davis. Working extensively with seeds, he would grind up what he was studying and test the resulting mush for things like proteins and enzymes to understand puzzles like what makes seeds sprout. This technique of analyzing tissue and looking at averages has been common throughout science, and theoretical frameworks were suited to the data it generated.

But seeds, Bradford began to see, also are individuals. Even among seeds from the same plant, some sprout early, with just a little water, while others wait, as if hedging their bets. Most sprout with a certain in-between amount of water – the classic bell curve of statistics.

“If all the seeds are ground up together, the information underlying that variation among individuals is lost,” Bradford said. Results at the far edges of the bell curve (or other graph) are considered “noise,” statistical “error” best ignored.

When Bradford started putting on his “population-colored glasses,” he began to see that biological variation is not random, but follows a predictable pattern. Each individual has its own pre-programmed threshold for responding to a stimulus. As that stimulus increases, more individuals cross their threshold and respond; and as the stimulus increases, their response speeds up. (Think of wildflowers bursting into bloom.) This threshold-based variation is a built-in, functional feature of biological systems, not a “bug.”

This mechanism brings efficiency to the system: When something happens – for example, a hormone arrives, or a pest – not every individual in the system has to respond. If the stimulus is a little bit, the low-threshold individuals respond first, rather than all individuals responding just a little bit.

It makes for a simple, automatic system that is flexible enough to respond to challenges in real time, yet stable.

However, much of science has been slow to see this. “It’s our blind spot,” Bradford argued. “We’re still stuck on the idea that, at the cellular or molecular level, the individuals are all identical.”