Quantitative Systems Pharmacology

Imagine having a "flight simulator" for a new medicine—a digital world where you could test its effectiveness, anticipate its side effects, and optimize its dosage for different types of patients, all before the first human trial. This capability would revolutionize the long, costly, and uncertain path of drug development. The fundamental challenge, however, is the immense complexity of the human body. How can we bridge the gap between a drug's action on a single molecule in a petri dish and its ultimate clinical outcome in a living, breathing person?

This is precisely the knowledge gap that Quantitative Systems Pharmacology (QSP) is designed to fill. QSP is a discipline that builds predictive, mathematical models of drugs, diseases, and patients by integrating our knowledge of physiology, pharmacology, and systems biology. It provides a rigorous framework for asking "what if" questions and understanding the dynamic interplay between a drug and the body.

This article will guide you through the world of QSP. First, we will explore the "Principles and Mechanisms," detailing how these models are constructed. We will see how they create a physiological map of the body to track a drug's journey and model the intricate biological circuitry that governs its effects. Following that, we will shift our focus to "Applications and Interdisciplinary Connections," discovering how these powerful virtual models are used to predict efficacy, ensure safety, unravel biological complexity, and move us toward the future of personalized medicine.

Imagine you are a detective tracking a mysterious agent through a bustling city. Your task is twofold. First, you need a map of the city—its highways, streets, districts, and barriers—to predict where your agent might travel, how quickly they move, and where they might get stuck or be removed from the scene. This is the geography of the problem. Second, you need to understand what the agent does when it reaches its destination. Does it deliver a message? Sabotage a piece of machinery? What is its mechanism of action? This is the operative part of the story.

Quantitative Systems Pharmacology (QSP) is this detective story, and the drug is our agent. It tackles the problem by beautifully merging two distinct but interconnected worlds: the mechanical "geography" of the whole body and the complex "biochemical circuitry" of disease at the cellular level. The first part is the domain of ​​Physiologically Based Pharmacokinetics (PBPK)​​, and the second is the core of ​​Quantitative Systems Pharmacology (QSP)​​. Let's embark on a journey to understand how these two perspectives unite to create a powerful predictive engine.

At first glance, the human body is an overwhelmingly complex system. But in physics and engineering, we have a powerful tradition: if you can't solve the whole complex problem at once, start with a simpler, but fundamentally correct, representation. For PBPK, this representation is a blueprint of the body as a hydraulic system.

The Blueprint and the Heartbeat

Think of the body as a collection of compartments—the liver, kidneys, brain, muscle, and so on—all connected by a network of pipes, the circulatory system. The heart is the master pump, generating a total flow of blood called the ​​cardiac output​​, or $Q_{CO}$. This total flow is then distributed among the different organs. If the flow to the liver is $Q_{liver}$ and the flow to the brain is $Q_{brain}$, then the principle of conservation of flow tells us something wonderfully simple: the sum of all the individual organ blood flows must equal the total cardiac output.

We can express the flow to any organ $i$, $Q_i$, as a fraction, $\phi_i$, of the cardiac output: $Q_i = \phi_i Q_{CO}$. These fractions, which are controlled by the body's vascular system, must all add up to one: $\sum \phi_i = 1$. This elegant constraint is the backbone of our physiological map. It ensures that our model of blood circulation is physically sound. If blood flow to the muscles increases during exercise (a higher $\phi_{muscle}$), the flow to other organs must decrease to compensate.

Now, with this plumbing in place, we can apply another fundamental law: ​​conservation of mass​​. For any organ, the rate at which the amount of drug accumulates is simply the rate it flows in minus the rate it flows out. The drug arrives via the arterial blood and leaves via the venous blood. This simple balance, written as a differential equation for each organ, forms the core of a PBPK model.

The Gatekeepers and the Unbound Drug

Of course, a drug molecule doesn't just passively ride the bloodstream. It interacts with its environment. One of the most important interactions is binding to proteins in the blood plasma. Think of these proteins as temporary "taxis." When a drug molecule is in a taxi, it is bound; when it's on its own, it is ​​unbound​​ or ​​free​​. This distinction is critical because of a central principle in pharmacology: the ​​unbound drug hypothesis​​. It states that only the unbound drug is free to leave the bloodstream, cross membranes, interact with its target, and be eliminated. The bound drug is just along for the ride.

We quantify this with the ​​unbound fraction​​, $f_u$, which is the ratio of the free concentration to the total concentration in the plasma ($f_{u,p}$). If $f_{u,p}$ is $0.01$, it means that at any given moment, $99\%$ of the drug in the blood is bound to proteins, and only $1\%$ is free to do its job.

This principle beautifully explains how drugs distribute into tissues. A drug's tendency to accumulate in a tissue depends on how it binds to components inside the tissue cells compared to how it binds to proteins in the blood. If we assume the system reaches a steady state where the unbound concentration inside the tissue, $C_{t, \text{unbound}}$, equals the unbound concentration in the plasma, $C_{p, \text{unbound}}$, we can derive a wonderfully simple and powerful relationship. The overall ​​tissue-to-plasma partition coefficient​​, $K_p$, which is the ratio of the total concentrations at steady state, is given by:

where $f_{u,t}$ is the unbound fraction within the tissue. This tells us that a drug will concentrate in tissues where it binds more tightly (lower $f_{u,t}$) relative to its binding in plasma.

Traffic Jams: Flow vs. Permeability

So, the free drug leaves the blood. But how fast? This depends on the nature of the "traffic jam." Is the delivery of drug to the tissue limited by the "highway" speed (blood flow, $Q$) or by the "toll booth" speed (the membrane's permeability, characterized by a ​​permeability-surface area product​​, $PS$)?

This leads to two essential types of tissue models:

• ​​Perfusion-Limited​​: For many small drugs and highly irrigated tissues, crossing the cell membrane is incredibly fast compared to the rate at which blood flow can deliver the drug. Here, $PS \gg Q$. The membrane is not a barrier; the rate-limiting step is perfusion (blood flow). In this case, we can assume the tissue and the blood leaving it are in instantaneous equilibrium. This simplifies the model immensely.
• ​​Permeability-Limited​​: For larger molecules or tissues with tight barriers (like the brain), getting across the membrane is the slow step. Here, $PS \ll Q$. No matter how fast blood delivers the drug, its uptake is gated by the barrier's low permeability. In this case, we must explicitly model the separate vascular and tissue compartments and the slow flux between them.

Understanding which regime a drug-tissue pair falls into is crucial for building a model that is both simple enough to be practical and complex enough to be accurate.

The Body's Filter: Metabolism and Clearance

Our detective story wouldn't be complete without the agent being removed from the scene. This process is called ​​clearance​​. The liver is the primary metabolic engine of the body, a sophisticated chemical plant that modifies foreign substances like drugs to facilitate their excretion.

To describe this, we introduce a concept called ​​intrinsic clearance​​, $CL_{int}$. This represents the raw, inherent metabolic capacity of the liver's enzymes for the unbound drug, independent of how fast the drug is delivered. It's a measure of the liver's "processing speed."

However, the liver can only metabolize the drug that is delivered to it by the blood flow, $Q$. The actual ​​hepatic clearance​​, $CL_h$, emerges from the interplay between blood flow and intrinsic clearance. A common way to model this is the ​​well-stirred model​​, which treats the liver as a single, perfectly mixed tank. This assumption leads to a clear and intuitive formula for the ​​extraction ratio​​, $E$—the fraction of drug removed in one pass through the liver:

The total hepatic clearance is then simply $CL_h = Q \times E$. This equation shows that if the intrinsic clearance is very high ($CL_{int} \gg Q$), $E$ approaches 1, and clearance is limited by blood flow ($CL_h \approx Q$). If the intrinsic clearance is very low ($CL_{int} \ll Q$), clearance depends mainly on the enzyme capacity and protein binding ($CL_h \approx f_u CL_{int}$).

But what if the enzyme system gets overwhelmed? Enzymes, like any machine, have a maximum operating speed, $V_{max}$. At low drug concentrations, the metabolic rate is proportional to the concentration. But at high concentrations, the enzymes become saturated, and the rate of metabolism hits a ceiling at $V_{max}$. This is described by the famous ​​Michaelis-Menten equation​​. When this happens, clearance is no longer a constant; it becomes dependent on the drug concentration. A higher dose leads to a lower effective clearance, causing the drug concentration to rise more than expected. This phenomenon of saturation is a gateway from the linear world of simple pharmacokinetics into the nonlinear, dynamic world of systems pharmacology.

The PBPK model gives us a precise prediction of the drug concentration over time at the site of action. But this is just the input to the real question: what does the drug do? This is where QSP takes the stage, modeling the target biological system as a dynamic network.

The Dynamics of Effect: Perturbing a Thermostat

Biological systems are rarely static; they are in a dynamic ​​steady state​​. Think of a thermostat maintaining a room's temperature. There's a constant production of heat (from the furnace) and a constant loss of heat (to the outside). The temperature is stable because these two rates are balanced.

Many biological responses, like the level of a key protein or biomarker $R$, can be described by a similar ​​turnover model​​:

Here, $k_{in}$ is the zero-order production rate (the furnace) and $k_{out}R$ is the first-order degradation rate (the heat loss). The baseline level of the biomarker, $R_0$, is simply $k_{in} / k_{out}$.

A drug typically doesn't invent a new biological function; it perturbs an existing one. It acts by changing the setting on the thermostat—by modulating either $k_{in}$ or $k_{out}$.

• ​​Inhibition of Production​​: A drug might block the synthesis of $R$. We can model this as $k_{in}' = k_{in}(1 - I(C))$, where $I(C)$ is an inhibitory function of the drug concentration $C$. This reduces the "heat" input, causing the "temperature" $R$ to drop.
• ​​Stimulation of Loss​​: A drug could enhance the degradation of $R$. We can model this as $k_{out}' = k_{out}(1 + S(C))$, where $S(C)$ is a stimulatory function. This increases the "heat" loss, also causing $R$ to drop.
• ​​Inhibition of Loss​​: Conversely, a drug could protect $R$ from degradation. This would be $k_{out}' = k_{out}(1 - I(C))$, reducing the "heat" loss and causing the "temperature" $R$ to rise.

This simple, elegant framework allows us to translate a drug's molecular action (inhibition or stimulation) into a dynamic, quantitative prediction of its effect on a biological system.

The Grand Unification: Coupled PBPK-QSP Models

Now we can unite our two worlds. The PBPK model predicts the drug concentration over time, which serves as the input driving the QSP model of the intracellular network. For a target tissue, this involves linking the drug concentration in the blood to that inside the cells.

For a tissue where diffusion across membranes is a rate-limiting step (a ​​permeability-limited model​​), this linkage can be described by a series of mass-balance equations. For example, the drug concentration in the interstitial space, $C_e(t)$, is governed by its exchange with the capillary plasma ($C_p$) and its uptake into the cells ($x_D$):

Here, $V_e$ is the volume of the interstitial space, while $PS_{cap}$ and $PS_{cell}$ are the permeability-surface area products for the capillary wall and the cell membrane, respectively. The plasma concentration $C_p$ is itself determined by a PBPK model describing the organ's vascular compartment.

Simultaneously, inside the cell, we have the systems biology model. This is a system of equations for all the intracellular species, which we can call the vector $x(t)$. The change in these species is due to the complex web of biochemical reactions, described by a ​​stoichiometric matrix​​ $S$ and a ​​reaction rate vector​​ $v(x, \theta)$, plus the influx of our drug from the interstitial space:

Here, $V_{cell}$ is the cell volume, $x_D$ is the intracellular drug concentration (one component of vector $x$), and the selector vector $e_D$ ensures the drug influx is added to the correct species. Solving these equations together provides a complete, multi-scale description of the drug's journey and its action. This is the mathematical embodiment of the QSP paradigm.

A model of an "average" human is useful, but the true power of QSP lies in its ability to predict how a population of diverse individuals will respond. After all, medicine is about treating people, not averages.

To do this, we must grapple with two concepts: ​​variability​​ and ​​uncertainty​​. In a Feynman-esque analogy, variability is the real, objective fact that a mountain range has peaks and valleys of different heights. Uncertainty is the fog obscuring our view, preventing us from knowing the exact height of any given peak. Variability is inherent to the population; uncertainty is a limitation of our knowledge. Variability is ​​aleatory​​ (due to chance, like a roll of the dice), while uncertainty is ​​epistemic​​ (due to a lack of knowledge, and can be reduced with more data).

QSP models embrace this by creating ​​virtual populations​​. Instead of one set of parameters, we generate thousands of sets, each representing a plausible virtual individual. This is not random guesswork. We use established physiological scaling laws (e.g., how organ sizes relate to body weight) and statistical distributions of real-world human characteristics (age, sex, weight, genetic markers). By combining these deterministic rules with statistical distributions for unexplained inter-individual differences, we create an in silico cohort that mirrors the diversity of a real clinical trial population. Running a simulation on this virtual population doesn't give us one answer; it gives us a distribution of answers, predicting who might respond well, who might not, and why.

With a complex model of a virtual population, a new question arises: which of the hundreds of parameters are the most important? Which biological knobs, when turned, have the biggest effect on the outcome? This is the job of ​​sensitivity analysis​​.

There are two main flavors. ​​Local sensitivity analysis​​ is like gently tapping one part of the machine while it's running at a specific setting. Mathematically, it involves calculating the partial derivatives of the model's output with respect to each parameter. It tells us the effect of small changes around a nominal value.

But biology is rarely so simple. Parameters can interact in nonlinear ways. The effect of changing parameter A might depend on the value of parameter B. To capture this, we need ​​global sensitivity analysis​​. This is like shaking the whole machine, varying all parameters across their entire plausible ranges simultaneously. Techniques like ​​Sobol indices​​ use a statistical approach (analysis of variance) to apportion the total uncertainty in the model's output to the uncertainty in each input parameter and their interactions. It tells us not just which knobs are sensitive, but which ones are driving the overall variability and which ones are involved in complex, synergistic effects.

This analysis is not just an academic exercise. It is the key to refining the model. It points us to the most influential biological pathways, telling us where to focus our experimental efforts to reduce the "fog" of uncertainty and build a more predictive and reliable model. It is how we learn from our virtual world to make better decisions in the real one.

Having journeyed through the fundamental principles of Quantitative Systems Pharmacology (QSP), we have learned the grammar and vocabulary of this exciting discipline. We have seen how differential equations can describe the intricate dance of molecules within our bodies. But to truly appreciate the power and beauty of QSP, we must see it in action. We now move from the "how" to the "what for." What marvels can we accomplish with this quantitative lens on biology?

QSP is far more than a set of equations; it is a mindset, a digital laboratory that empowers us to ask, and answer, some of the most challenging "what if" questions in medicine. It serves as a bridge, connecting the controlled world of the laboratory bench to the complex, dynamic reality of the patient's bedside. It is, in essence, a flight simulator for drug development, allowing us to test, refine, and predict the performance of a new medicine long before it reaches a human being. Let us explore the vast landscape of its applications.

At the heart of drug development lies a fundamental duality: we want a drug to be effective against a disease, but we also need it to be safe for the patient. This creates a "therapeutic window" of doses—high enough to work, low enough to be tolerated. QSP is the essential tool for mapping out this window.

Will the Drug Work? From Basic Clearance to Complex Immunotherapies

Before a drug can exert its therapeutic effect, it must survive a perilous journey through the body. One of the most formidable gatekeepers on this journey is the liver, a magnificent biochemical processing plant that clears foreign substances from the blood. How much of our drug will survive this passage? A QSP model can tell us. By integrating physiological facts, like the rate of blood flow to the liver ($Q_h$), with drug-specific properties, like its intrinsic tendency to be metabolized ($CL_{int}$) and its penchant for binding to proteins in the blood ($f_u$), we can construct a "well-stirred" model of the liver. This simple yet powerful model predicts the hepatic clearance ($CL_h$), a critical parameter that dictates how long the drug stays in the body and at what concentration. It reveals fascinating, non-intuitive behaviors: for certain drugs, changing protein binding has a dramatic effect on clearance, while for others, it has almost none at all. Understanding these rules is fundamental to designing a drug that reaches its target at the right concentration for the right amount of time.

But what about drugs that don't just act, but orchestrate? Consider a modern marvel of immuno-oncology: the Bispecific T-cell Engager, or BiTE. This is not a blunt instrument but a molecular matchmaker. One arm of the BiTE grabs onto a tumor cell, and the other grabs a passing T-cell, the soldier of our immune system. By forcing this introduction, the BiTE unleashes the T-cell's killing power directly against the cancer. The question of efficacy is no longer just about drug concentration; it's about the dynamics of a three-way interaction between drug, cancer, and the immune system.

Here, QSP truly shines. We can build a virtual patient, a system of equations that simulates this entire battle. We model the BiTE's journey through the bloodstream, its arrival at the tumor, the fraction of T-cells it activates, the subsequent cascade of signaling molecules (cytokines) they release, and ultimately, the rate at which the tumor shrinks. This simulation allows us to explore which biological knobs are most important for success. Is it the drug's half-life? Its binding strength? The patient's initial number of T-cells? By performing a "sensitivity analysis," we can identify the most critical parameters, guiding the design of more potent and effective cancer therapies.

Will the Drug Be Safe? From Targeted Toxicities to Systemic Storms

Efficacy is only half the story. The other, equally important, half is safety. The same BiTE that so beautifully orchestrates the killing of tumor cells carries a dark-side risk. What if the antigen it targets on the tumor is also present, even at low levels, on healthy, vital cells in the heart or lungs? Our molecular matchmaker might inadvertently introduce the immune system's assassins to innocent bystanders. This is known as "on-target, off-tumor" toxicity, a major challenge in targeted therapy. QSP provides the tools to quantify this risk. By modeling the activation probability of a T-cell based on the number of antigens on a cell's surface, we can calculate the drug concentration that is high enough to kill tumor cells (which have many antigens) but low enough to spare normal cells (which have few). This defines a quantitative "selectivity margin," a critical guide for safe dosing.

Another major safety concern arises from drug-drug interactions (DDIs). A patient with multiple conditions might take a cocktail of medications. What if one drug for blood pressure interferes with the liver enzymes responsible for clearing another drug for cholesterol? The concentration of the cholesterol drug could build up to toxic levels, with disastrous consequences. QSP allows us to build dynamic models of these interactions. We can simulate the pharmacokinetics of an "inhibitor" drug and model its effect on the activity of a key metabolic enzyme over time. We can then predict how this will impact the clearance and exposure of a "victim" drug co-administered with it. These simulations act as a virtual drug interaction screening tool, helping to prevent dangerous surprises in the clinic.

Sometimes, a drug's toxicity is not about a specific organ but about a system-wide overreaction. Potent immunotherapies can sometimes trigger a "cytokine storm," a dangerous, runaway inflammatory response. Predicting the risk of such an event is incredibly complex, as it depends on the drug's potency, where it distributes in the body, and a patient's individual inflammatory state. A QSP-based risk model can integrate these disparate factors. It can model the drug's concentration in different tissues, calculate the resulting receptor activation in each, and weigh their contributions to a systemic hazard. By further incorporating a patient-specific "susceptibility factor," the model can generate a personalized probability of an adverse event, turning a vague risk into a quantifiable number.

The body's response to a drug is rarely a simple, linear affair. It is a symphony of dynamic, interconnected, and regulated processes. QSP gives us the power to look beyond simple cause-and-effect and understand the rich, complex music of physiology.

The Dance of Time: Why Effect Doesn't Always Follow Concentration

We often assume that as a drug's concentration in the blood rises and falls, its effect does the same in perfect lockstep. But biology is often much slower and more subtle. Consider an inhaled corticosteroid used to treat airway inflammation. The drug may enter and leave the plasma quickly, but its effect on airway remodeling can linger for much longer. This is because the drug must first find its target receptors within the lung tissue, bind to them, and then initiate a cascade of gene expression changes. These are not instantaneous processes.

If we plot the drug's effect against its plasma concentration over time, we don't get a simple line; we get a loop, a phenomenon known as hysteresis. QSP models can perfectly explain this. By explicitly modeling the slow "on" and "off" rates of the drug binding to its tissue receptors, we can reproduce this lag between pharmacokinetics (PK) and pharmacodynamics (PD). Understanding this hysteresis is crucial for determining the correct dosing schedule—it tells us that the effect can persist long after the drug concentration has peaked.

The Body as a Control System: Homeostasis and Stability

The body is a master of self-regulation. Through a vast network of feedback loops, it maintains a stable internal environment, a state known as homeostasis. For example, the Renin-Angiotensin-Aldosterone System (RAAS) is a beautiful hormonal cascade that regulates blood pressure. When we introduce a drug, such as an ACE inhibitor to treat hypertension, we are not just pushing on a single lever; we are intervening in a complex, closed-loop control system. The system will react, pushing back in ways that can either help or hinder the therapy.

QSP allows us to view physiology through the lens of control theory. We can map out the RAAS as a circuit diagram, with its activators, inhibitors, and feedback pathways. Then, by linearizing the system of equations around its steady state, we can analyze its stability, much like an engineer would analyze the stability of a bridge or an aircraft. The eigenvalues of the system's Jacobian matrix tell us about its fundamental modes of response—how quickly it returns to balance after being disturbed. This analysis can reveal how a drug changes the system's stability, providing deep insights into its mechanism and potential for both therapeutic benefit and adverse feedback-driven effects.

New Frontiers: The Rational Design of Vaccines

The power of QSP extends beyond traditional therapeutics to the design of vaccines. Imagine we are developing a next-generation vaccine using a "single-cycle" viral vector. This clever vector can infect a cell and produce the antigen that trains our immune system, but it cannot produce new infectious viruses to spread further. This constrains the infection, which is a key safety feature. However, within each infected cell, the vector's genome can be engineered to undergo multiple rounds of amplification, producing a larger "burst" of antigen from each single infection event.

This presents a classic engineering trade-off. More antigen per cell could lead to a stronger, more protective immune response at a lower vaccine dose (improved efficacy). But it could also trigger a stronger innate inflammatory reaction, leading to more side effects like fever and fatigue (reduced safety). QSP allows us to build a model that captures this trade-off explicitly. We can define mathematical relationships for both the probability of protection and the systemic hazard, both as a function of the vaccine dose and its intracellular amplification factor. By analyzing this model, we can find the optimal level of amplification that maximizes the safety margin, guiding the rational, model-informed design of a better, safer vaccine.

The "average human" is a statistical fiction. We are all unique, and our responses to medicine vary accordingly. The ultimate promise of QSP is to move beyond one-size-fits-all medicine and towards a future where treatments are tailored to the individual.

Special Populations: Beyond the 'Standard' Human

A pregnant woman's body is a world undergoing constant and profound change. Her plasma volume expands, her cardiac output increases, and her metabolism shifts. A drug dose that is safe and effective for a non-pregnant adult might be ineffective or dangerous during pregnancy. Treating such "special populations" requires a deep understanding of how these physiological changes impact a drug's journey through the body.

PBPK models, a cornerstone of QSP, are perfectly suited for this challenge. We can create a "dynamic avatar" of a pregnant patient, where physiological parameters like blood volume and organ blood flow are not fixed constants but functions of time, changing as the pregnancy progresses. By simulating a drug's behavior in this dynamic model—for instance, a therapeutic monoclonal antibody that relies on a recycling mechanism called FcRn to prolong its half-life—we can predict how its exposure and target engagement will change through each trimester. This allows for dose adjustments that ensure the therapy remains safe and effective for both mother and child.

Pharmacogenomics: Reading the Individual's Instruction Manual

Perhaps the most exciting frontier is personalized medicine based on our own unique genetic code. Our DNA is our personal instruction manual, and subtle variations in it can have significant effects on how we respond to drugs. Why does the same dose of a medication produce a powerful therapeutic effect in one person, a mild effect in another, and unpleasant side effects in a third?

QSP provides a framework to connect the dots from genotype to phenotype. Consider the deeply personal response to a serotonergic psychedelic. It is known that this experience is mediated by the serotonin 2A ($5\text{-HT}_{2A}$) receptor. A QSP model can begin with an individual's specific genetic variations (polymorphisms) in the gene that codes for this receptor. These variations can be translated into quantitative changes in the receptor's properties: one variant might increase the number of receptors expressed on a neuron's surface, while another might alter its binding affinity for the drug, and a third might change its intrinsic signaling efficacy.

The model integrates these genetic effects to calculate a personalized, effective signal strength for a given drug dose. This signal is then calibrated to a standardized score representing the predicted intensity of the subjective experience. This is a breathtaking example of QSP in action: weaving together genomics, molecular biology, and pharmacology to make a quantitative prediction about a deeply personal human experience.

In conclusion, Quantitative Systems Pharmacology is a computational crucible where we can forge our understanding of biology and medicine. It is a discipline that unifies diverse fields, allowing us to build models that are not just descriptive, but predictive. By creating these virtual worlds, these digital patients, we can design safer and more effective drugs, unravel the complexities of disease, and take meaningful steps toward the ultimate goal of true precision medicine for every individual.