Based on Wikipedia: In silico

In 1987, a scientist at Los Alamos National Laboratory coined a phrase that would come to describe one of the most profound shifts in how we do science. Christopher Langton was announcing a workshop on artificial life, and he needed a term for experiments that happened entirely inside computers. He called them "in silico."

It was a joke. A pun, really.

For centuries, scientists had used Latin phrases to describe where their experiments took place. "In vivo" meant inside a living organism—testing a drug on a mouse, watching bacteria multiply in a petri dish implanted in a rabbit. "In vitro" meant in glass, referring to the test tubes and beakers where reactions happened outside living bodies. "In situ" meant studying something exactly where you found it in nature.

Langton's addition was playful pseudo-Latin: "in silicon," a nod to the silicon chips that powered the computers now hosting a new kind of experiment. The phrase caught on immediately, perhaps because it captured something genuinely new about science in the computer age.

From Joke to Revolution

Two years after Langton's workshop, a Mexican mathematician named Pedro Miramontes presented something remarkable at a cellular automata conference in Los Alamos. He had run experiments on DNA and RNA entirely inside a computer—no test tubes, no living organisms, just mathematical models of molecular behavior. It was among the first true "in silico" biology experiments, and Miramontes would later expand this work into his doctoral dissertation.

The timing wasn't coincidental. By the late 1980s, computers had finally become powerful enough to simulate biological processes in meaningful ways. Not perfectly—not even close to perfectly—but well enough to ask interesting questions and sometimes get useful answers.

What makes in silico work different from simply "using a computer"? The distinction matters. When a scientist uses a spreadsheet to analyze data from a lab experiment, that's not in silico research. When they use a computer to statistically process results from clinical trials, that's not in silico either. In silico means the experiment itself—the thing being observed, manipulated, and measured—exists only as a simulation.

Think of it this way: in vivo experiments involve actual living systems, in vitro experiments involve real molecules in real containers, but in silico experiments involve mathematical representations running as software. The "organism" being studied is made of code.

Why This Matters for Drug Discovery

Developing a new medicine is staggeringly expensive. A pharmaceutical company might spend ten to fifteen years and several billion dollars to bring a single drug to market. Much of that time and money goes to experiments that fail.

Consider the traditional approach to finding a drug candidate. You have a disease, and you know which protein in the body is causing problems. You need a molecule that will attach to that protein and block its harmful activity. How do you find such a molecule?

One method is called high-throughput screening. You build a robotic laboratory that can physically test thousands of different chemical compounds every day. Each compound gets mixed with the target protein to see if it sticks. The hit rate is typically around one percent or less—and most of those "hits" will fail in subsequent testing.

Now imagine doing this in silico instead. You create a detailed three-dimensional model of your target protein, accurate down to the position of individual atoms. Then you use algorithms to simulate how millions of different molecules would interact with that protein. The computer can test virtual compounds far faster than any robot, and far more cheaply than synthesizing real chemicals.

In 2010, researchers demonstrated how powerful this approach could be. Using a protein docking algorithm called EADock, they searched for molecules that might inhibit an enzyme associated with cancer. The algorithm ranked candidates by how well they fit into the enzyme's active site—like finding keys that might fit a particular lock, except the lock is a protein and the keys are potential drugs.

When they tested the top candidates in actual lab experiments, fifty percent of them worked. Half. Compare that to the one percent hit rate of traditional screening. The in silico approach had found genuine leads fifty times more efficiently.

The COVID-19 Moment

When SARS-CoV-2 emerged in late 2019, scientists around the world faced an urgent question: could any existing drugs be repurposed to fight this new virus?

Testing every approved medication in clinical trials would take years. There wasn't time. So researchers turned to in silico methods, using computer models of the virus's key proteins to screen thousands of existing drugs for potential activity. They were looking for molecules that might interfere with how the virus entered cells or replicated inside them.

This kind of "drug repurposing" study became a major application of in silico techniques during the pandemic. Rather than starting from scratch to develop new compounds, scientists could virtually test medications already proven safe in humans. If the computer predicted a match, that drug could move quickly into clinical trials.

The approach didn't produce miracle cures, but it dramatically accelerated the search for treatments. What might have taken years of laboratory screening could be narrowed down in weeks of computation.

Simulating Life Itself

Drug discovery is just one application. Some researchers have more ambitious goals: building complete computer models of living cells.

In 2007, a team created an in silico model of Mycobacterium tuberculosis, the bacterium that causes TB. Tuberculosis is famously slow-growing—studying its behavior in the lab means waiting months to observe what happens. But in the computer model, those months could be compressed into minutes. Researchers could watch virtual bacteria respond to virtual drugs in simulated fast-forward.

Other projects have modeled specific cellular processes, like the growth cycle of Caulobacter crescentus, a bacterium often used in laboratory research. These models capture how cells divide, how they regulate their internal chemistry, how they respond to their environment.

But here's the humbling reality: we're nowhere close to simulating an entire cell with full accuracy.

A single bacterial cell contains thousands of different proteins, each folding into precise three-dimensional shapes, each interacting with dozens of other molecules in carefully choreographed chemical dances. The number of simultaneous reactions happening inside even the simplest cell is staggering. Our understanding of how all these pieces work together remains fragmentary, and even if we understood perfectly, we lack the computing power to simulate it all.

Current in silico cell models require massive simplifying assumptions. They capture some behaviors well while ignoring others entirely. They're useful tools, not faithful reproductions. The map is not the territory, and our maps of cellular life remain crude sketches.

The Bioinformatics Revolution

There's another sense in which biology has moved into silicon: genetic information itself has become digital.

When scientists sequence DNA, they transform the physical molecule into a string of letters—A, T, G, and C, representing the four nucleotide bases. These sequences can be stored in databases, searched, compared, and analyzed using software. The genetic code has become, quite literally, computer code.

Researchers can now design genes on their computers and then use artificial gene synthesis to create the actual DNA molecules. The workflow runs backward from traditional biology: instead of studying natural genes to understand their sequences, scientists write sequences and manufacture the corresponding genes.

This is in silico biology at its most literal. The boundary between information and life has become permeable. What exists first as data can be rendered into molecules, and what exists as molecules can be captured as data.

The Broader Applications

In silico methods have spread far beyond drug discovery and cellular modeling. They're now used to:

• Analyze entire cellular systems in organisms from E. coli bacteria to human cell lines
• Optimize industrial bioprocesses—figuring out how to get microorganisms to produce more of a desired compound
• Simulate clinical trials before running expensive real-world studies
• Design proteins with specific functions using software like RosettaDesign
• Interpret the flood of data coming from genome sequencing, gene expression studies, and protein analysis

The European Grid Infrastructure has been used to run massive simulated oncology trials, distributing the computation across networks of computers to achieve processing power that single machines can't match. Projects like Folding@home harness the spare computing cycles of volunteers' home computers to simulate how proteins fold—a problem so computationally intensive that it benefits from millions of machines working together.

Silicon and Living Tissue

The boundaries keep blurring. "Organ-on-a-chip" technology creates tiny devices where actual living cells are arranged to mimic the structure and function of human organs. These chips are physical objects—not simulations—but they're designed to interface with electronic sensors and computational analysis. They sit somewhere between in vitro and in silico, biological systems intimately coupled with silicon.

Meanwhile, "in silico medicine" has emerged as a field unto itself, focused on using computational models to predict how individual patients will respond to treatments. Rather than relying on averages from clinical trials, doctors might someday simulate your specific physiology to determine which drug, at which dose, will work best for you.

The Fundamental Shift

When Langton coined "in silico" in 1987, he was describing something genuinely unprecedented. For all of human history, experiments had required physical stuff—organisms, chemicals, materials you could touch and measure. Now experiments could happen in a realm of pure information.

This doesn't replace traditional biology. Simulations must be validated against reality. Drugs that look promising in silico still need to be tested in vitro and in vivo before reaching patients. The computer model is a tool for generating hypotheses and narrowing searches, not a substitute for empirical verification.

But in silico methods have fundamentally changed what's possible. They let us explore vast spaces of possibilities that no laboratory could ever physically test. They let us observe processes too fast, too slow, too small, or too dangerous to study directly. They let us ask "what if?" questions that would be impractical or unethical to investigate with living systems.

The pseudo-Latin joke has become a pillar of modern science. Biology, chemistry, medicine, and pharmacology now routinely split their work between the laboratory bench and the computer cluster. The silicon has become as essential as the test tube.

And as computing power continues its exponential growth, the experiments we can run in silico will only become more sophisticated, more accurate, and more central to how we understand life itself.