{
  "id": "proteomics",
  "title": "Proteomics",
  "lang": "en",
  "source_lang": "en",
  "summary": "Proteomics is the large-scale study of proteins, including their abundance, molecular forms, interactions, locations, and biological functions.",
  "category": "science",
  "subcategory": "biology",
  "type": "concept",
  "aliases": [
    "Protein science at proteome scale"
  ],
  "generated_at": "2026-11-04",
  "model": "pro/gpt-6.1-sol",
  "url": "https://aiwiki.page/en/proteomics/",
  "translations": {
    "zh-Hans": {
      "url": "https://aiwiki.page/zh/proteomics/",
      "json": "https://aiwiki.page/zh/proteomics.json"
    }
  },
  "sections": [
    {
      "heading": "",
      "text": "Proteomics is the systematic, large-scale study of proteins in a biological system. Its subject is the proteome: the collection of proteins present in an organism, tissue, cell, compartment, or sample under specified conditions. Unlike a genome sequence, a proteome changes substantially with development, environment, physiological state, and disease. Proteomics therefore investigates not only which proteins are present, but also their quantities, molecular modifications, locations, interactions, and rates of production and removal.\n\nProteins carry out much of the molecular work of life, acting as enzymes, receptors, transporters, structural components, and regulators. Their properties cannot be inferred completely from DNA or RNA measurements. Proteomics connects genetic information to cellular mechanisms and observable biological states, complementing genomics, transcriptomics, and other large-scale approaches. Its principal technologies include mass spectrometry, separation methods, affinity-based assays, and computational analysis."
    },
    {
      "heading": "The proteome as a dynamic biological system",
      "text": "A proteome is defined by its biological boundaries and sampling conditions. The proteome of a cultured bacterial population differs conceptually from that of human blood plasma, an isolated organelle, or a developing tissue. Even genetically similar cells may contain different protein complements because they occupy different developmental states or respond differently to their surroundings. A reported proteome is also an experimental observation: it includes what a method detects, rather than necessarily every protein physically present.\n\nThe relationship between genome and proteome is not one-to-one. Human cells possess roughly 20,000 protein-coding genes, but a single gene can generate multiple RNA transcripts through alternative splicing, and those transcripts may produce different protein sequences. Sequence variants, alternative translation initiation, cleavage, and chemical modification further diversify the products. A proteoform is a particular molecular form of a protein arising from a specific sequence and combination of modifications. Identifying a gene product does not, by itself, establish which proteoforms exist or what functions they perform.\n\nProtein abundance depends on synthesis, transport, dilution through cell division, and degradation. Messenger RNA abundance influences synthesis but does not determine protein concentration uniquely: translation efficiency and protein lifetime vary widely. In a simple model, protein abundance \\(P\\) changes according to\n\n\\[\\frac{dP}{dt}=k_{\\mathrm{synthesis}}-k_{\\mathrm{loss}}P.\\]\n\nAt steady state, this gives \\(P=k_{\\mathrm{synthesis}}/k_{\\mathrm{loss}}\\). The model illustrates why equal transcript levels can coexist with different protein abundances, although real cells often require additional terms for compartmental transport, growth, and regulated degradation. Measurements made during stress or differentiation may also capture systems far from steady state.\n\nProteomics distinguishes abundance from activity. A signaling enzyme may become active through phosphorylation without any substantial change in its total quantity. A receptor may move from an internal compartment to the cell membrane, changing cellular responsiveness while its overall abundance remains constant. An enzyme may be abundant but inactive because a required cofactor is absent. Consequently, expression proteomics, modification analysis, localization studies, and activity measurements answer related but different questions. A complete functional account often requires several of them together, along with measurements of metabolites and physiological behavior."
    },
    {
      "heading": "Historical development and the emergence of large-scale analysis",
      "text": "Proteomics grew out of protein chemistry, biochemical separation, and analytical instrumentation rather than from a single invention. Early protein research typically isolated one protein or a small group before measuring composition, catalytic activity, or structure. Such work established the chemical basis of protein function but was difficult to extend systematically to thousands of gene products. Advances in chromatography and gel electrophoresis gradually made complex mixtures accessible to comparative analysis.\n\nA major step was high-resolution two-dimensional gel electrophoresis. In 1975, Patrick O’Farrell and, independently, Joachim Klose published influential methods that separated proteins in two stages: first by isoelectric point and then largely by molecular size. A single gel could resolve hundreds or thousands of spots, allowing comparisons between biological samples. However, a spot was not necessarily a single molecular species, and detecting a spot did not automatically reveal its sequence. Very hydrophobic, scarce, or unusually large or small proteins were also difficult to represent comprehensively.\n\nThe introduction of soft ionization methods transformed mass spectrometric protein analysis. Electrospray ionization, developed for biomolecular analysis through work associated with John Fenn and colleagues, transfers molecules from solution into gas-phase ions, often with multiple charges. Matrix-assisted laser desorption/ionization, developed during the 1980s, uses a light-absorbing matrix to facilitate ion formation. These approaches made large biomolecules and complex peptide mixtures substantially more tractable. Fenn and Koichi Tanaka shared part of the 2002 Nobel Prize in Chemistry for soft desorption ionization methods used in biological macromolecule analysis.\n\nThe term “proteome,” generally attributed to Marc Wilkins in 1994, expressed the ambition to study proteins on a scale analogous to the genome. Expanding sequence databases were crucial: peptide masses and fragmentation patterns could increasingly be interpreted by comparison with predicted sequences rather than by laborious manual sequencing. During the late 1990s and early 2000s, liquid chromatography coupled to tandem mass spectrometry established peptide-mixture analysis as a central route to large-scale protein identification.\n\nSubsequent progress involved both instruments and experimental design. Stable-isotope labeling, isobaric chemical tags, higher-resolution analyzers, faster fragmentation cycles, and improved statistical methods made reproducible quantitative comparisons possible. International initiatives, including the Human Proteome Organization, founded in 2001, promoted coordinated research and reporting standards. The Human Proteome Project, launched in 2010, extended this effort toward documenting human protein existence and biological roles, while distinguishing strong experimental evidence from uncertain assignments."
    },
    {
      "heading": "Sample preparation and molecular separation",
      "text": "Proteomic measurements begin with decisions about what material to collect and what biological question the sample can answer. Tissue composition, collection time, preservation, extraction chemistry, and handling delays can all change the resulting protein profile. A tumor biopsy, for example, contains not only malignant cells but also immune cells, blood vessels, connective tissue, and other components. Differences between two bulk samples may therefore reflect changing cell proportions rather than altered regulation within a single cell type.\n\nExtraction must release proteins while limiting unwanted chemical and enzymatic changes. Detergents and chaotropic reagents disrupt membranes and protein structure; inhibitors can reduce proteolysis or loss of phosphorylation during processing. Proteins are often reduced to break disulfide bonds, followed by alkylation to limit their reformation. These procedures improve accessibility and analytical consistency, but preparation itself can introduce artifacts, including oxidation or modifications caused by prolonged exposure to particular reagents. Membrane proteins and insoluble structural proteins require especially careful extraction because conditions suitable for soluble proteins may recover them poorly.\n\nIn bottom-up proteomics, proteins are digested into peptides before measurement. Trypsin is widely used because it generally cleaves after lysine or arginine, except in certain sequence contexts, producing peptides well suited to many mass spectrometric workflows. Other proteases provide complementary sequence coverage. Digestion reduces molecular complexity in one sense—large proteins become analytically manageable peptides—but increases the number of distinct analytes. Incomplete cleavage, variable recovery, and peptide instability can introduce differences unrelated to the original biological abundance.\n\nSeparation reduces the number of species entering an instrument simultaneously. Reversed-phase liquid chromatography, commonly operated at low flow rates for sensitive analyses, separates peptides largely by hydrophobic interactions with the stationary phase. Additional fractionation by charge, size, or high-pH reversed-phase chromatography can increase coverage, although it also increases instrument time and opportunities for sample loss. Two-dimensional gels remain useful for visualizing certain protein patterns and changes in apparent size or charge, but they no longer dominate large-scale discovery workflows.\n\nSpecialized enrichment is often necessary for proteins or modifications present at low abundance. Phosphopeptides can be enriched with immobilized metal affinity materials or metal oxides; glycopeptides require other chemical or affinity strategies. Depleting abundant plasma proteins can reveal less abundant components, but may also remove proteins bound to those abundant carriers. Preparation therefore imposes a selective window on the proteome. Recovery controls, reference materials, randomized processing order, and repeated quality-control samples help distinguish biological variation from technical effects introduced before measurement."
    },
    {
      "heading": "Mass spectrometric identification and acquisition strategies",
      "text": "Mass spectrometry measures gas-phase ions according to their mass-to-charge ratio, conventionally written \\(m/z\\). In a typical bottom-up experiment, chromatography delivers peptides to an ion source, an initial mass spectrum records peptide-ion signals, and selected ions undergo fragmentation. Tandem mass spectrometry measures the resulting fragments, whose mass differences provide information about peptide sequence. Accurate identification depends on interpreting the precursor mass, fragment pattern, charge state, and chemical modifications together.\n\nMany instruments combine components with different roles. Quadrupoles can isolate ions within selected \\(m/z\\) ranges; time-of-flight analyzers infer mass-to-charge ratios from ion flight behavior; Orbitrap analyzers measure frequencies associated with trapped-ion motion. Instrument resolution, mass accuracy, sensitivity, and acquisition speed influence which peptides are distinguished and how many can be analyzed during a chromatographic peak. Higher performance does not remove the problem of overlapping signals, particularly in samples containing many components at very different concentrations.\n\nIn data-dependent acquisition, the instrument repeatedly surveys the mixture and selects particular precursor ions—often among the most intense—for fragmentation. This provides interpretable spectra and strong identification performance but can miss low-abundance peptides or select different precursors in repeated runs. Data-independent acquisition instead fragments ions systematically across predefined or adaptively arranged mass windows. Because many precursors may contribute to each fragmentation spectrum, computational methods must separate their overlapping signals using chromatographic behavior and predicted or experimentally measured fragment patterns.\n\nTargeted proteomics concentrates measurement effort on predefined peptides. Selected reaction monitoring, often performed with triple-quadrupole instruments, monitors specified precursor-to-fragment transitions. Parallel reaction monitoring records multiple fragments from selected precursors using a high-resolution analyzer. These approaches can support consistent measurement of a limited panel across many samples, but assay specificity still depends on peptide selection, interference assessment, and appropriate standards. A peptide suitable for identification is not automatically suitable for robust quantification in a difficult biological matrix.\n\nBottom-up analysis usually identifies proteins indirectly from their peptides. Top-down proteomics instead measures intact proteins and their fragments, preserving connections among sequence variations and modifications that digestion can obscure. It is particularly valuable for distinguishing proteoforms, but intact-protein separation, fragmentation, and spectral interpretation become more difficult as molecular size and mixture complexity increase. Middle-down approaches analyze larger fragments than conventional bottom-up workflows. The choice among them reflects a trade-off between breadth of coverage, molecular detail, sensitivity, and analytical practicality rather than a universally superior method."
    },
    {
      "heading": "Quantification, experimental design, and measurement confidence",
      "text": "Quantitative proteomics estimates differences in protein or peptide abundance between samples, or determines amounts relative to calibrated standards. Label-free workflows commonly integrate chromatographic peptide-ion intensities across runs. These measurements are influenced by peptide recovery, ionization efficiency, instrument response, and interference. Intensities from different peptides therefore cannot generally be compared as if equal signals represented equal molar amounts. Comparing the same peptide across well-controlled samples is usually more defensible than comparing unrelated peptide signals directly.\n\nStable-isotope methods create chemically similar analytes that can be distinguished by mass. In stable isotope labeling by amino acids in cell culture, cells incorporate labeled amino acids into proteins, allowing differently labeled populations to be combined before much of the analytical processing. Chemical labeling can be applied after extraction or digestion. Isobaric tags, including tandem mass tags, allow several samples to be combined and distinguished through reporter ions generated during fragmentation. However, co-isolated peptides can contaminate reporter measurements and compress apparent abundance ratios; narrower isolation, additional fragmentation stages, or computational approaches can reduce particular forms of interference.\n\nAbsolute quantification requires a defined connection between signal and amount. Synthetic peptides containing stable isotopes can serve as internal standards, enabling endogenous and labeled peptide signals to be compared. The point at which a standard enters the workflow matters: a peptide added after digestion does not directly control for extraction loss or incomplete digestion. Full-length labeled protein standards can account for more processing steps, although producing and characterizing them is more demanding. Calibration, linearity, limits of quantification, matrix effects, and measurement uncertainty remain essential even when instruments produce highly precise numerical outputs.\n\nBiological replication and technical replication answer different questions. Repeated injections of one digest characterize aspects of analytical repeatability; independently collected biological samples reveal variation among organisms, cultures, or tissues. Randomization and balanced processing reduce confounding between experimental condition and analytical batch. Normalization can correct systematic differences in sample loading or response, but assumptions such as an unchanged majority of proteins may fail when a treatment causes broad proteome remodeling.\n\nMissing values are another central issue. A peptide may be absent biologically, below detection, excluded during precursor selection, or lost during processing. Replacing every missing measurement with an arbitrary low value can generate misleading differences. Statistical models must account for the acquisition strategy and plausible missingness mechanism. Results are most informative when effect sizes, uncertainty, replication, and multiple-testing procedures are reported together, rather than when a list is defined only by an arbitrary fold-change threshold."
    },
    {
      "heading": "Functional, spatial, and interaction proteomics",
      "text": "Protein abundance maps describe composition, but biological function also depends on chemical state, molecular partners, and location. Post-translational modifications include phosphorylation, acetylation, methylation, ubiquitination, and glycosylation. These changes can regulate catalytic activity, stability, trafficking, or molecular recognition. Modification-focused experiments usually enrich modified peptides before analysis because their signals may otherwise be overwhelmed by unmodified molecules. Identifying a modified peptide also differs from determining exactly which residue carries the modification.\n\nChanges in modification abundance require careful interpretation. An increased phosphopeptide signal may result from more total protein, a greater fraction of molecules phosphorylated at that site, or both. Site occupancy—the fraction of relevant protein molecules carrying a modification—is not equivalent to modified-peptide intensity. Estimating occupancy may require measurement of modified and unmodified forms, protein abundance, and additional experimental constraints. Modifications can also occur in combinations, creating regulatory states that conventional short-peptide measurements do not always reconstruct.\n\nProtein–protein interactions are studied using affinity purification followed by mass spectrometry, among other methods. A tagged protein or an antibody is used to isolate a molecular assembly, and associated proteins are identified. Appropriate controls are needed because abundant proteins and proteins that bind nonspecifically to beads or tags can appear as apparent partners. Moreover, co-purification establishes association under the experimental conditions, not necessarily direct physical contact between every pair of recovered proteins.\n\nProximity labeling addresses a related but distinct question. An enzyme positioned near a protein or within a compartment chemically labels nearby molecules, which are subsequently enriched and identified. Such methods can capture transient or spatially restricted neighborhoods, but proximity is not synonymous with binding. Cross-linking mass spectrometry provides another type of evidence: chemical cross-linkers join reactive groups within a limited spatial range, and identification of linked peptides constrains possible molecular arrangements. These constraints can complement cryo-electron microscopy and X-ray crystallography in structural studies.\n\nSpatial proteomics maps proteins to organelles, subcellular fractions, or tissue regions. Fractionation profiles can distinguish proteins associated with the mitochondrion, endoplasmic reticulum, or other compartments, while imaging and microdissection preserve aspects of anatomical context. Activity-based profiling uses reactive probes to label accessible functional sites in particular enzyme families. Thermal proteome profiling measures changes in protein solubility after heating and can reveal responses consistent with ligand binding or altered molecular interactions. Each method samples a different functional property; none provides a complete or automatically causal account of protein behavior."
    },
    {
      "heading": "Computational inference, databases, and reproducibility",
      "text": "Bioinformatics connects instrument signals to biological interpretation. In a conventional database-search workflow, software generates theoretical peptides from reference protein sequences, predicts their possible fragment masses, and compares them with measured spectra. Search parameters specify digestion rules, allowed modifications, precursor and fragment tolerances, and other assumptions. Expanding the search to include many variants or modifications can recover additional biology, but also enlarges the space of candidate explanations and makes statistical discrimination more demanding.\n\nIdentification confidence is often assessed through a target–decoy strategy. Spectra are searched against authentic sequences and artificial alternatives designed to represent incorrect matches. The observed decoy matches help estimate the false discovery rate among accepted results. A commonly used threshold is 1%, but its meaning depends on the level of analysis: spectrum matches, distinct peptides, protein groups, and modification sites require separate consideration. An estimated 1% false discovery rate is not a statement that every accepted identification has a 99% probability of being correct.\n\nProtein inference is difficult because different proteins can contain identical peptides. Closely related gene products and splice isoforms may therefore be indistinguishable unless uniquely informative peptides are detected. Software may report a protein group rather than a single resolved sequence. Strong claims about a previously undetected protein, a sequence variant, or an unusual translation product require especially careful examination of uniqueness, alternative explanations, and spectral quality. A reference database is itself a scientific choice: sample-specific variants or microbial sequences may be missing, while very large databases can introduce additional ambiguity.\n\nModern workflows increasingly use machine learning to predict peptide fragmentation, chromatographic retention, and ion mobility, or to improve match scoring. These predictions support data-independent acquisition analysis and can reduce dependence on experimentally acquired spectral libraries. Their performance must nevertheless be evaluated under the conditions in which they are applied. A model trained mainly on common unmodified tryptic peptides may perform differently on unusual modifications, alternative proteases, or unfamiliar instrumentation. Signal transfer between runs also requires error control rather than an assumption that similar retention time proves identity.\n\nReproducibility depends on access to raw data, metadata, processing settings, and code or sufficiently detailed software configurations. Public repositories such as PRIDE and MassIVE participate in the ProteomeXchange framework, which facilitates deposition and discovery of proteomics datasets. Community reporting practices distinguish raw observations from processed abundance tables and biological interpretation. Batch effects, sample misidentification, contaminants, and inconsistent annotation can otherwise survive into apparently polished results. Pathway enrichment and network analyses are useful for organizing findings, but depend on database coverage and do not independently establish mechanism or causation."
    },
    {
      "heading": "Biological applications, clinical translation, and emerging scales",
      "text": "Proteomics is widely used to investigate cellular responses, developmental transitions, host–pathogen interactions, and disease mechanisms. In cancer research, combining protein measurements with genome and RNA data can reveal whether a genetic alteration is accompanied by changes in protein abundance, signaling, or molecular complexes. Such proteogenomic analysis is valuable because sequence changes do not propagate uniformly through every molecular layer. It can also show that different genetic alterations converge on similar protein-level processes.\n\nClinical applications range from measuring established protein analytes to discovering candidate biomarkers and characterizing disease-associated molecular patterns. Mass spectrometric identification of microbial isolates by their protein profiles is an established diagnostic application of matrix-assisted laser desorption/ionization. Mass spectrometry can also support the typing of amyloid deposits in specialized diagnostic workflows. These applications should be distinguished from exploratory studies in which a protein signature separates small research groups but has not yet demonstrated generalizable clinical performance.\n\nBlood-based discovery presents particular challenges. A small number of abundant proteins dominate plasma protein mass, while many tissue-derived proteins occur at far lower concentrations. Age, medication, kidney function, inflammation, collection procedures, and storage can influence measured profiles. Affinity platforms using antibodies or aptamers can measure large panels with comparatively high throughput, but their signals depend on reagent binding and may be affected by cross-reactivity or sequence variation at binding sites. Agreement with mass spectrometry or other independent methods is therefore an important form of analytical corroboration. Clinical translation additionally requires defined intended use, independent validation, and evidence appropriate to the diagnostic claim.\n\nEnvironmental and microbial community studies apply metaproteomics to identify proteins expressed by mixed populations. Together with metagenomics, this can connect community composition to functions such as nutrient acquisition or carbon transformation. Interpretation is complicated by shared peptide sequences among related organisms and by incomplete reference databases. In agriculture and industrial biotechnology, proteomics helps investigate crop stress responses, microbial production systems, and changes in enzymes involved in metabolism. In drug research, chemical proteomics can identify candidate binding partners, unintended targets, and downstream responses without assuming that the intended target explains every observed effect.\n\nSingle-cell proteomics and low-input tissue methods extend analysis to biological variation hidden by bulk averaging. Improvements in miniaturized preparation, low-flow separations, ion handling, and multiplexing have made it possible to measure many proteins from individual cells, although coverage depends strongly on cell size, instrument, and workflow. Protein measurements cannot exploit a general amplification process equivalent to polymerase chain reaction, making sample loss and sensitivity especially consequential. Spatially resolved sampling further links protein composition to tissue architecture. The resulting datasets require careful separation of genuine cellular heterogeneity from differences in recovery, missing measurements, and analytical sensitivity."
    }
  ],
  "links": [
    "protein",
    "cell",
    "enzyme",
    "dna",
    "mass-spectrometry",
    "genome",
    "gene",
    "alternative-splicing",
    "messenger-rna",
    "cell-membrane",
    "cofactor",
    "chromatography",
    "gel-electrophoresis",
    "electrospray-ionization",
    "nobel-prize-in-chemistry",
    "disulfide-bond",
    "ion",
    "tandem-mass-spectrometry",
    "amino-acid",
    "isotope",
    "calibration",
    "measurement-uncertainty",
    "post-translational-modification",
    "glycosylation",
    "protein-protein-interaction",
    "antibody",
    "cryo-electron-microscopy",
    "x-ray-crystallography",
    "mitochondrion",
    "endoplasmic-reticulum",
    "bioinformatics",
    "false-discovery-rate",
    "machine-learning",
    "causation",
    "cancer",
    "biomarker",
    "metagenomics",
    "metabolism",
    "polymerase-chain-reaction"
  ],
  "sources": [],
  "license": "https://creativecommons.org/licenses/by/4.0/"
}