Scientific glossary, A to Z

If 2 of 100 people have an event within one year, their one-year absolute risk is 2%; the group and time period are essential context. (Cochrane: interpreting results)

A decrease from 4% to 3% is an absolute reduction of 1 percentage point, even though the relative reduction is 25%. (Cochrane: interpreting results)

Absorption is one part of pharmacokinetics and should not be confused with the eventual biological effect of a drug. (Merck Manual: pharmacokinetics)

Approval through this pathway is still approval, but confirmatory research must establish the anticipated clinical benefit. (FDA: accelerated approval)

A test can give closely repeated results yet still be inaccurate if they are consistently shifted away from the reference. (FDA / ICH: analytical validation)

Acetylation is one type of acylation. Its effects depend on the attachment site and molecule; it does not universally improve biological activity. (Review: peptide and protein therapeutics; Review: peptide hormone modifications)

An enzyme lowers this barrier, allowing a reaction to proceed faster without changing its overall free-energy difference. (OpenStax: enzymes)

A molecule occupying the active site can prevent the normal substrate from binding. (OpenStax: enzymes)

Acetylation adds a small acyl group; fatty acylation adds a longer fatty-acid-derived group. These are related modifications, not interchangeable names for one change. (Review: peptide hormone modifications; Review: peptide and protein therapeutics)

Reporting an adverse event raises a safety question but does not, by itself, establish that the product caused the event. (FDA: adverse events)

Binding and activating are different: both an agonist and an antagonist can have affinity for the same receptor. (Merck Manual: drug-receptor interactions; Review: ligand binding and response models)

Not every aggregate is an amyloid fibril. Aggregation can affect a preparation's quality and biological behavior, so it is distinct from the folding of a single molecule. (Review: peptide and protein therapeutics)

Think of a receptor as a switch: an agonist can turn it on, although the size of the response depends on the biological system. (Merck Manual: drug-receptor interactions)

Albumin is a carrier, whereas an enzyme is a catalyst; proteins can have very different jobs. (OpenStax: proteins)

This prevents foreknowledge from influencing who enters each group and differs from blinding after allocation. (Cochrane: bias in randomized trials)

Binding away from the active site can alter the protein's shape and either increase or decrease activity. (OpenStax: enzymes)

Picture a spiral staircase, with the backbone forming the spiral and the side chains pointing outward. (OpenStax: proteins)

A sequence ending in -CONH2 is chemically different from the corresponding free carboxyl terminus. The change can affect charge, stability and activity, depending on the peptide. (Review: peptide and protein therapeutics)

A peptide bond is one kind of amide linkage, while a terminal amide is a modification at a chain end. (IUPAC: peptide terminology)

The side chain distinguishes one standard amino acid from another, helping determine its chemical behavior. (OpenStax: proteins)

Amyloid describes a structural form of aggregated protein, not a single molecule or one particular disease. (MedlinePlus: amyloidosis; Review: amyloid fibril structure)

Adding a known amount to a sample can help assess recovery. A result near 100 percent in one matrix does not establish performance in every matrix or concentration range. (FDA / ICH: analytical validation)

Impurities, degradation products and the sample matrix can interfere, so specificity must be evaluated rather than assumed. (FDA / ICH: analytical validation)

Validation of an identity test is not identical to validation of a quantitative impurity assay, because their intended uses differ. (FDA / ICH: analytical validation)

Blocking a signal is not always the same as lowering every cell function; the result depends on what the blocked signal normally does. (Merck Manual: drug-receptor interactions)

Always check the interval and analyte. AUC from dosing to the last measurable sample is not automatically the same as AUC extrapolated to infinity. (FDA: pharmacokinetic bioequivalence endpoints)

Ask what the assay actually measures: an identity test, an amount measurement and an activity assay answer different questions. (FDA / ICH: analytical validation)

Autocrine signaling differs from endocrine signaling, where the signal travels through the circulation to distant targets. (OpenStax: cell signaling)

The amino-acid side chains extend from this framework, so changing a side chain is different from breaking the backbone. (OpenStax: proteins)

Unlike an alpha-helix, a beta-sheet resembles a folded sheet rather than a coil. (OpenStax: proteins)

Unlike random scatter, bias can consistently favor one answer, such as when the groups being compared differ in important ways. (Cochrane: bias in randomized trials)

Absorption and bioavailability are related but not identical, because some absorbed drug can be metabolized before reaching the systemic circulation. (Merck Manual: bioavailability)

For many systemically acting products, studies compare exposure measures such as AUC and Cmax using prespecified criteria. Similar chemical identity alone does not establish bioequivalence. (FDA: pharmacokinetic bioequivalence endpoints)

Not every biologic is simply a protein made in a cell; this category also includes products such as blood components and some gene therapies. (FDA: drug approval glossary)

A biomarker is not automatically a validated surrogate for clinical benefit; its role depends on the evidence and context of use. (FDA-NIH: BEST glossary)

Blinding participants, care providers and outcome assessors addresses different potential biases, so a paper should say who was blinded. (Cochrane: bias in randomized trials)

A case report can highlight an unusual observation, but it lacks a comparison group needed to estimate a treatment's effect reliably. (NCI: case report; Cochrane: bias in randomized trials)

The signaling molecule can stay outside the cell while the receptor relays information inward. (OpenStax: cell signaling)

A narrow confidence interval does not eliminate concerns about bias, inconsistent results or whether the evidence applies to the intended population. (Cochrane: interpreting results)

A useful COA identifies the batch, tests, limits and results; it is not a blanket guarantee that a product is safe for any use. (FDA / ICH: manufacturing and quality)

A sample label identifies a specimen, while a chain-of-custody record shows who controlled it and when. This helps establish sample integrity and accountability. (EPA: sample receipt and chain of custody)

Your left and right hands illustrate the idea; most standard amino acids are chiral, but glycine is not. (OpenStax: amino-acid structures)

Different interactions with the two phases can separate a mixture into distinct peaks. (IUPAC: chromatography nomenclature)

Clearance is expressed as a volume per time, not simply as the percentage of a dose that disappears. (Merck Manual: pharmacokinetics)

Choose the actual endpoint when interpreting a trial: symptom change and survival are not interchangeable outcomes. (FDA-NIH: BEST glossary)

A tiny change may have a small p-value in a very large study while offering little meaningful benefit. (Cochrane: interpreting results)

Cmax depends on the sampling schedule as well as the drug and administration conditions. A sparse schedule may miss the true peak. (FDA: pharmacokinetic bioequivalence endpoints)

Coenzymes differ from inorganic helpers such as metal ions, although both support enzyme function. (OpenStax: enzymes)

Magnesium and zinc can act as enzyme helpers; the enzyme's amino-acid chain is not always sufficient for its function. (OpenStax: enzymes)

A cohort can be followed forward or studied using existing records; because exposure is not randomly assigned, confounding remains a concern. (Cochrane: non-randomized studies)

Under a simple reversible competitive model, raising substrate concentration can reduce the inhibitor's apparent effect. (OpenStax: enzymes; NCATS: enzymatic assays)

Compounded drugs are not FDA-approved products, and FDA does not review their safety, effectiveness or quality before marketing. (FDA: compounding questions)

A 95% confidence interval does not mean there is a 95% probability that this particular interval contains the true value; its coverage refers to the method across repeated studies. (Cochrane: interpreting results)

If groups differ in age and age affects the outcome, an apparent treatment difference may partly reflect age rather than the treatment. (Cochrane: non-randomized studies)

A control may receive a placebo, usual care, an active treatment or no intervention, depending on the research question. (Cochrane: research glossary)

In single-particle cryo-EM, many particle images are combined computationally to reconstruct a three-dimensional map. (RCSB PDB: determining structures)

A signal at a cell-surface receptor can lead to a change in cAMP inside the cell, helping relay the message. (OpenStax: cell signaling)

Cyclization can constrain shape, but a ring alone does not establish stability, oral absorption, safety or clinical effectiveness. (Review: peptide and protein therapeutics)

D and L describe configuration, not positive or negative optical rotation. Glycine has no D/L pair because its alpha carbon is not chiral. (OpenStax: amino-acid structures)

A kilodalton, abbreviated kDa, is 1,000 daltons; this unit describes molecular mass rather than the mass of a whole sample. (IUPAC: atomic mass unit)

A degradation product differs from a residual starting material: one forms through change of the drug substance, while the other can remain from manufacturing. (FDA / ICH: degradation products)

Cooking an egg illustrates a visible change in protein structure; denaturation is different from cutting a protein into shorter chains. (OpenStax: proteins)

The prefix di- means two; adding a third residue gives a tripeptide. (IUPAC: peptide terminology)

Under the simple equilibrium binding model, free ligand equal to Kd gives half occupancy. Lower Kd indicates tighter binding under the same conditions; Kd is not automatically EC50. (Review: ligand binding and response models)

The concentration measured in blood does not automatically equal the concentration at every tissue target. (Merck Manual: pharmacokinetics)

Disulfide bonds can connect different parts of one chain or join separate chains, helping stabilize a protein's shape. (OpenStax: proteins)

The order of its bases carries biological information, including instructions used to make RNA and proteins. (NHGRI: genetics glossary)

A dose-response graph can reveal potency, the maximum response and how steeply response changes with dose. (Merck Manual: dose-response relationships)

The label alone is incomplete: check whether the treating clinicians, outcome assessors and analysts were also blinded. (Cochrane: bias in randomized trials)

A low EC50 describes potency in that assay, not proof of a larger maximum response or better benefit in patients. (NCATS: relative EC50)

Each cycle reveals the next residue and leaves a shorter chain for the next cycle; longer proteins may first be broken into smaller fragments. (OpenStax / LibreTexts: Edman sequencing)

A small p-value does not tell you whether the effect itself is large or important; read the effect estimate and its uncertainty. (Cochrane: interpreting results)

Two drugs can differ in potency while reaching the same maximum response; clinical efficacy uses a related but outcome-focused meaning. (Merck Manual: dose-response relationships)

Two enantiomers can have different interactions with biological molecules even though their atoms are connected in the same order. (OpenStax: amino-acid structures)

The bloodstream carries the message, while cells need the appropriate receptors to respond. (OpenStax: cell signaling)

Killing bacteria does not necessarily remove their endotoxin, so sterility testing and endotoxin testing answer different questions. (FDA: bacterial endotoxins)

Most enzymes are proteins, but some RNA molecules also catalyze reactions; an enzyme does not make an unfavorable reaction favorable by changing its overall free-energy difference. (OpenStax: enzymes)

The label inactive does not mean irrelevant to every patient: dyes, preservatives and other formulation ingredients can still matter clinically. (FDA: inactive ingredients; FDA: compounding questions)

Excretion and metabolism are different processes: one removes material, while the other changes its chemical form. (Merck Manual: pharmacokinetics)

Evidence from one age group or disease setting may not transfer directly to another, even when the original study is well conducted. (Cochrane: interpreting results)

In the Beer-Lambert relationship, absorbance depends on this coefficient, concentration and the distance light travels through the sample. (IUPAC: Beer-Lambert law)

A registered facility, a submitted IND or a COA is not the same thing as approval of a particular drug product. (FDA: drug approval glossary)

The product acts like a stop signal, slowing additional production when enough has accumulated. (OpenStax: enzymes)

A drug can be absorbed from the gut yet have reduced systemic availability because of first-pass metabolism. Absorption and bioavailability are therefore not synonyms. (Merck Manual: bioavailability)

Freeze-drying can help stabilize suitable formulations, but it does not itself make a contaminated preparation sterile. (FDA: lyophilization)

In GPCR signaling, receptor activation promotes GDP-GTP exchange, enabling the G protein to influence downstream targets. (OpenStax: signaling pathways)

The G protein can then influence an enzyme or ion channel, connecting an outside signal to an intracellular response. (OpenStax: cell signaling)

Not every gene encodes a protein; some genes produce RNA that functions without being translated. (NHGRI: genetics glossary)

A gene can be present in a cell's DNA without being expressed at the same level in every cell or circumstance. (NHGRI: genetics glossary)

Attachment to an asparagine nitrogen is called N-linked glycosylation; attachment through a hydroxyl oxygen is O-linked. Glycosylation differs from nonenzymatic glycation. (Review: protein glycosylation)

A GMP claim concerns manufacturing controls; it is not the same as FDA approval of a product or proof of clinical benefit. Requirements depend on the product and jurisdiction. (FDA / ICH: manufacturing and quality)

GRADE considers risk of bias, inconsistency, indirectness, imprecision and publication bias; it is not the same as VALREXA's own evidence labels. (Cochrane: interpreting results)

For a drug, the elimination half-life describes concentration decline, not automatically how long a clinical effect lasts. (Merck Manual: pharmacokinetics; StatPearls: pharmacokinetics)

When pH equals pKa, the corresponding acid and conjugate-base concentrations are equal under the equation's assumptions. (OpenStax / LibreTexts: amino-acid acid-base chemistry)

Studies asking superficially similar questions may still differ enough that pooling their results requires caution. (Cochrane: research glossary)

An HPLC result can help assess composition, but what it establishes depends on the method, detector, standards and validation. (IUPAC: chromatography nomenclature; EJPPS: chromatographic impurity analysis)

Hydrogen bonds help stabilize alpha-helices and beta-sheets; they are different from the covalent peptide bonds joining the backbone. (OpenStax: proteins)

Protease-catalyzed cleavage of a peptide bond is a form of hydrolysis. Hydrolysis can also affect other bond types and need not be enzyme-catalyzed. (OpenStax: synthesis and hydrolysis)

Many polar or charged side chains tend to face water on the surface of a soluble protein. (OpenStax: proteins)

Nonpolar side chains often cluster inside soluble proteins, although membrane proteins face a different environment. (OpenStax: proteins)

In many soluble proteins, hydrophobic side chains are buried inside while more water-friendly groups face outward. (OpenStax: proteins)

An IC50 can change with substrate concentration and assay design, so numbers from different experiments are not automatically comparable. (NCATS: enzymatic assays)

An immune response may or may not change treatment effect or cause harm. Assessment depends on the product, assays and clinical context, not simply peptide size. (Review: peptide and protein therapeutics)

The applicable definition depends on the material. Excipients intentionally added to a drug product are not automatically impurities, and different impurities need different analytical methods. (FDA / ICH: manufacturing and quality)

An effect in a dish is not the same as evidence that an intervention benefits people. (NCI: in vitro)

An in vivo study may involve an animal rather than a person, so the species and study design still matter. (NCI: in vivo)

Signing a form documents part of the process but does not replace explanation or opportunities for questions. Consent information may need updating during a study. (ClinicalTrials.gov: research glossary)

IRB review considers issues such as risks, informed consent and participant protections. Approval does not prove that an intervention works. (ClinicalTrials.gov: research glossary)

Retaining assigned groups helps preserve randomization. Missing outcome data remain a potential source of bias and require appropriate handling and transparent reporting. (Cochrane: bias in randomized trials)

An IND allowing a study to proceed is not FDA approval to market the product or proof that the treatment works. (FDA: clinical research and INDs)

Zero net charge does not mean every group is uncharged; positive and negative charges can balance one another. (OpenStax / LibreTexts: amino-acid acid-base chemistry)

Protein kinases can add phosphate groups to particular amino-acid side chains as part of cell signaling. (OpenStax: cell signaling)

Km depends on the enzyme, substrate and conditions, and should not automatically be treated as a direct measure of binding affinity. (NCATS: enzymatic assays)

L does not mean that a molecule rotates polarized light to the left. Glycine is achiral and is not assigned D or L. (OpenStax: amino-acid structures)

Chromatography separates components before the mass spectrometer examines the eluting material, providing complementary information. (Penn State / IUPAC: mass spectrometry)

Ligand is a broad binding term: some ligands activate receptors, while others block or modify their responses. (Merck Manual: drug-receptor interactions)

Detected does not necessarily mean accurately quantified; the limit of quantitation addresses that separate question. (FDA / ICH: analytical validation)

A trace signal below the LOQ may indicate something is present without supporting a reliable numerical concentration. (FDA / ICH: analytical validation)

A lot number connects a sample to its manufacturing and testing records; a COA for one lot does not automatically describe another. (FDA / ICH: manufacturing and quality)

Mass data can support molecular identification, but interpretation also depends on factors such as charge state, fragmentation and other analytical evidence. (Penn State / IUPAC: mass spectrometry)

Uncertainty is not the same as a known error. State how it was evaluated and what an interval or coverage factor means rather than reporting an unexplained plus-or-minus number. (NIST: expressing measurement uncertainty)

Combining results does not erase bias, and a pooled estimate is only as interpretable as the included studies and methods. (Cochrane: research glossary)

Drug metabolism is one part of the absorption, distribution, metabolism and excretion processes that shape drug exposure. (Merck Manual: pharmacokinetics)

Robustness studies examine sensitivity to controlled changes, such as a small change in chromatographic conditions. This differs from repeating the same procedure without deliberate changes. (FDA / ICH: analytical validation)

A monograph can specify identity tests, strength and impurity limits; meeting a quality standard is different from proving benefit for a medical use. (USP: monographs)

A mutation can affect a protein or gene regulation, but not every sequence change causes disease. (NHGRI: genetics glossary)

These names describe chain direction even when an end is chemically modified, such as an acetylated N-terminus or amidated C-terminus. (IUPAC: peptide terminology)

Submitting an application is not the same as receiving approval; the FDA reviews the supporting product and manufacturing evidence. (FDA: drug approval glossary)

Failing to reject a null hypothesis does not prove it true. Interpretation also needs effect estimates, confidence intervals and the study's design limitations. (Cochrane: interpreting results; StatPearls: statistical power and errors)

If risk falls from 10% to 5%, the absolute reduction is 0.05 and the NNT for one fewer event is 1/0.05 = 20, not a guarantee for each group of 20. (Cochrane: interpreting results)

Cohort and case-control studies are examples; observed differences may reflect confounding as well as the exposure itself. (Cochrane: research glossary)

Odds and risks are not the same, so an odds ratio should not automatically be read as a risk ratio, especially for common outcomes. (Cochrane: effect measures)

A p-value is not the probability that the hypothesis is true and does not measure the size or clinical importance of an effect. (Cochrane: interpreting results)

The signal travels locally through the space between cells rather than primarily through the bloodstream to distant organs. (OpenStax: cell signaling)

A partial agonist may also reduce a full agonist's effect by competing for receptor occupancy. (Merck Manual: drug-receptor interactions)

Peer review can improve a paper but is not a guarantee that the findings are correct or will replicate. (NLM: peer-reviewed literature)

There is no universal peptide-versus-protein cutoff, so the familiar 50-residue boundary is a convention rather than a rule of nature. (IUPAC: peptide terminology)

In the usual backbone linkage, the carboxyl carbon of one residue is connected to the amino nitrogen of the next. (IUPAC: peptide terminology; OpenStax: peptide-bond chemistry)

Receptor activation and the resulting physiological response are pharmacodynamic questions; movement of the drug through the body is pharmacokinetics. (Merck Manual: pharmacodynamics)

Absorption, distribution, metabolism and excretion are often grouped as ADME; they help explain why concentration changes after administration. (Merck Manual: pharmacokinetics)

Kinases add phosphate groups and phosphatases remove them, providing complementary ways to regulate signaling. (OpenStax: cell signaling)

Breaking down cAMP can help terminate an intracellular signal. (OpenStax: cell signaling)

Phosphorylation can activate some proteins and inhibit others; it is not a universal on switch. (OpenStax: signaling pathways)

When pH is equal to that group's pKa, its corresponding protonated and deprotonated forms are present in equal amounts under the usual approximation. (OpenStax / LibreTexts: amino-acid acid-base chemistry)

A placebo-controlled study compares outcomes between groups; improvements within the placebo group alone do not isolate a placebo effect from natural change or other influences. (NCI: placebo; Cochrane: bias in randomized trials)

Polypeptide describes a chain; a functional protein can contain one such chain or multiple chains assembled together. (IUPAC: peptide terminology; OpenStax: proteins)

Examples include adding a phosphate group or cutting away part of a precursor chain; the final functional product can differ from the initial chain. (OpenStax: proteins)

A more potent drug can produce a specified response at a lower dose, but that does not automatically mean a greater maximum effect or better clinical benefit. (Merck Manual: dose-response relationships)

Precision describes scatter, not correctness; tightly grouped results can still be biased away from the accepted value. (FDA / ICH: analytical validation)

A preprint allows early access, but its findings may change after critique, revision or journal review. (NIH: preprints)

Preregistration separates planned tests from later exploratory decisions. Changes can be justified and documented; registration does not itself establish study quality. (Center for Open Science: preregistration)

Check how and when the outcome was measured and whether it matches the registered protocol. A study may specify more than one primary outcome. (ClinicalTrials.gov: research glossary)

This is the sequence level of description, before considering local helices, sheets or the chain's full three-dimensional fold. (OpenStax: proteins)

Hemoglobin transports oxygen, while an enzyme catalyzes a reaction; proteins are not all hormones or all enzymes. (OpenStax: proteins)

A PDB entry is a model supported by experimental data, not a guarantee that every atom's position is equally certain. (RCSB PDB: determining structures)

Sequence, chemical interactions and sometimes helper proteins called chaperones contribute to folding. (OpenStax: proteins; RCSB PDB: protein disorder and flexibility)

Proteolysis can activate a precursor, recycle a protein or destroy a signaling peptide. It is not always accidental damage. (Review: peptide and protein therapeutics)

If positive studies are easier to find than negative ones, a review can overestimate benefit even when it uses published studies carefully. (Cochrane: research glossary)

A high main-peak percentage is not automatically the same as active-ingredient content, correct identity, absence of endotoxin or sterility. (EJPPS: chromatographic impurity analysis; FDA / ICH: analytical validation; FDA: bacterial endotoxins)

A single-chain protein has no multi-subunit arrangement to describe at this level. (OpenStax: proteins)

Randomization helps balance confounders on average, but does not by itself fix missing data, poor measurement or other sources of bias. (Cochrane: bias in randomized trials)

A receptor is one part of a signaling system; what happens after binding depends on the ligand and the downstream machinery. (OpenStax: cell signaling)

A secondary standard can be qualified by comparison with a primary standard. A material's label alone does not establish its identity, purity or suitability for a particular method. (FDA / ICH: manufacturing and quality)

If risk is 2% versus 4%, the risk ratio is 0.5, while the absolute difference is 2 percentage points; both views help interpret the result. (Cochrane: interpreting results)

Replication asks whether a finding is supported again. Re-running the same code on the same data instead addresses computational reproducibility; terminology can vary across disciplines. (Center for Open Science: replication)

A chain containing ten amino-acid units has ten residues; the term refers to the linked units rather than free amino acids. (IUPAC: peptide terminology)

A matching retention time can support an identification, but it is not conclusive on its own because different substances can elute together. (IUPAC: chromatography nomenclature)

Messenger RNA provides instructions for protein synthesis, while other RNA types have structural, regulatory or catalytic roles. (NHGRI: genetics glossary)

Ten measurements from each of ten participants do not automatically provide the same independent information as measurements from one hundred participants. (Cochrane: interpreting results; StatPearls: statistical power and errors)

Cyclic AMP and calcium ions are examples, carrying information inside the cell rather than serving as the original outside signal. (OpenStax: cell signaling)

Secondary results can be useful, but interpretation should consider the analysis plan, multiple comparisons and whether the study was designed to answer that question. (ClinicalTrials.gov: research glossary)

Secondary structure describes recurring shapes within a chain, not the complete fold of the whole protein. (OpenStax: proteins)

Selectivity is relative, not a promise that a drug affects only one receptor or has no off-target effects. (Merck Manual: drug-receptor interactions)

Order matters, not just composition: two chains containing the same kinds and numbers of amino acids can have different sequences. (IUPAC: peptide terminology; OpenStax: proteins)

Serious is a regulatory category based on outcomes and medical importance, rather than a simple description of how intense a symptom feels. (FDA: adverse events)

Its chemistry helps determine whether the residue is charged, polar or nonpolar and how it behaves within a protein. (OpenStax: proteins)

Receptor binding can begin a cascade involving enzymes and second messengers before the final cellular effect occurs. (OpenStax: cell signaling)

In the conventional approach, the C-terminal end is attached to a resin and protected amino acids are added toward the N-terminal end. (Nobel Prize: solid-phase synthesis)

A single purity percentage is not a complete specification. Identity, relevant impurities and other attributes may each require separate acceptance criteria. (FDA / ICH: manufacturing and quality)

A stability study supports storage conditions and a retest or expiry period; the result applies to the tested material and conditions. (FDA / ICH: manufacturing and quality)

Power depends on factors including sample size, effect size, variability and the significance threshold. Low power can leave an important effect undetected. (StatPearls: statistical power and errors)

Crossing a threshold does not establish a large effect or clinical importance; the estimate, uncertainty and study limitations still matter. (Cochrane: interpreting results)

Steady state during intermittent administration still includes peaks and troughs. It does not mean that concentration is flat at every moment. (StatPearls: pharmacokinetics)

A sterility test samples material and does not prove the absence of every other contaminant, including bacterial endotoxin. (21 CFR: sterility definition; FDA: bacterial endotoxins)

For an enzyme that breaks a molecule apart, that molecule is the substrate and the resulting pieces are products. (OpenStax: enzymes)

A laboratory marker may be useful, but improving that marker does not automatically establish an improvement in the clinical outcome it is meant to predict. (FDA: accelerated approval)

A chromatographic system may have criteria for peak resolution or repeatability. Passing these checks is not a substitute for validating the method or confirming sample identity. (FDA / ICH: analytical validation)

Some systematic reviews include a meta-analysis, while others do not because their evidence cannot sensibly be pooled. (Cochrane: research glossary)

Interactions among side chains help bring distant parts of the sequence together into a folded shape. (OpenStax: proteins)

Some settings use a TD50/ED50 dose ratio; others compare concentrations. It is a population-based comparison, not a guarantee that an individual exposure is safe. (Merck Manual: dose-response relationships; Open RN: pharmacokinetics and pharmacodynamics)

The window is a range, whereas the therapeutic index is a ratio. Values depend on the drug, endpoint and patient population and are not a self-dosing guide. (Open RN: pharmacokinetics and pharmacodynamics)

For an amino acid, the curve can reveal separate proton-loss steps and buffering regions associated with different pKa values. (OpenStax / LibreTexts: amino-acid acid-base chemistry)

Tmax describes timing, not the peak concentration itself or the duration of a clinical effect. (FDA: pharmacokinetic bioequivalence endpoints)

Transcription produces an RNA copy, whereas translation uses an RNA message to build a protein chain. (NHGRI: genetics glossary)

Translation connects a nucleotide message to a peptide sequence; it is distinct from copying DNA into RNA. (NHGRI: genetics glossary)

Count the linked amino-acid units rather than the number of peptide bonds when identifying chain length. (IUPAC: peptide terminology)

Vmax depends on how much active enzyme is present, so comparing values requires attention to enzyme amount and assay conditions. (NCATS: enzymatic assays)

This is a calculation describing distribution, not a literal anatomical space that must fit inside the body. (Merck Manual: pharmacokinetics)

The diffraction data support an electron-density map, which scientists interpret to place atoms; flexible or poorly resolved regions can remain uncertain. (RCSB PDB: determining structures)

An amino acid can have a positively charged amino group and a negatively charged carboxyl group at the same time. (OpenStax / LibreTexts: amino-acid acid-base chemistry)