A genome is data. Precision is a decision system.
Sequence alone does not determine care. Precision medicine connects sample provenance, molecular variation, phenotype, family history, population evidence, clinical context, patient goals and measurable utility into a decision that can be explained, acted on and revised.
IDENTITY
Identity, specimen type, collection conditions, consent, handling and clinical context determine whether downstream molecular results belong to the right person and question.
Genomics describes variation. Precision medicine governs its use.
Genomics examines genomes and their function at individual or population scale. Clinical genetics focuses on inherited or acquired variation relevant to health. Precision medicine integrates molecular information with non-genomic evidence to guide prevention, diagnosis or treatment. These fields overlap, but a sequence result is not automatically a diagnosis, prognosis or treatment recommendation.
Person, specimen, collection, handling, identity and consent remain linked.
PROVENANCEPlatform, coverage, quality, reference build, pipeline and version are explicit.
MEASUREObserved molecular difference is described without overclaiming clinical meaning.
CALLPopulation, computational, functional, segregation and clinical sources are evaluated.
INTERPRETPhenotype, family history, ancestry, treatment, environment and preferences frame relevance.
INTEGRATEAction, owner, threshold, uncertainty, alternatives and review trigger are defined.
GOVERNEvery transformation adds information—and creates a new place for error.
Define the clinical question, patient context, test scope, possible outcomes and consent needs.
Collect the correct specimen with verified identity, quality and chain of custody.
Produce molecular data with documented platform, controls, coverage and limitations.
Evaluate variants against curated evidence, phenotype, inheritance and uncertainty.
Connect the result to a clinically useful choice, counseling and accountable follow-through.
Reassess when evidence, phenotype, family information or available interventions change.
The right test is the smallest scope that can answer the clinical question with acceptable uncertainty.
Known locus or limited gene set
High depth and focused interpretation when phenotype or family evidence sharply constrains the question.
Defined condition architecture
Balances breadth, coverage and interpretation burden for genetically heterogeneous disorders.
Protein-coding regions
Broad discovery across coding sequence, with uneven coverage and limited detection of many noncoding or structural events.
Broad molecular survey
Captures more variant classes and noncoding sequence but expands interpretation, infrastructure and incidental findings.
More sequence does not guarantee more useful knowledge. Test choice must consider variant classes, expected diagnostic yield, turnaround, coverage, consent, interpretive capacity and consequence of uncertain findings.
A variant classification is a structured evidence judgment—not a property stored permanently in the DNA.
| Evidence domain | Question | Useful evidence | Common failure | Control |
|---|---|---|---|---|
| Population frequency | Is the variant compatible with disease prevalence and inheritance? | Representative reference populations, quality filters and ancestry context | Aggregate frequency hides subgroup or technical artifacts | Review population structure, coverage and disease model |
| Computational prediction | Does sequence or structure suggest functional impact? | Concordant, calibrated tools used as supporting evidence | Multiple correlated tools counted as independent proof | Weight modestly and seek orthogonal evidence |
| Functional evidence | Does the variant alter a disease-relevant function? | Validated assay with controls, biological relevance and quantitative effect | Artificial system or assay signal overgeneralized to disease | Evaluate assay validity and mechanism relevance |
| Segregation / de novo | Does inheritance support the proposed disease relationship? | Verified family relationships, phenotype and informative meioses | Small family or incomplete penetrance interpreted as certainty | Model penetrance, phenocopies and ascertainment |
| Clinical observations | Are affected individuals consistently linked to the variant? | Well-phenotyped cases, case–control data and independent replication | Duplicate reports or weak phenotype matching inflate evidence | Deduplicate, curate and compare alternative explanations |
| Gene–disease validity | Is the gene itself credibly related to the condition? | Human genetics, mechanism, replication and disease-specific evidence | Variant interpreted before gene–disease relationship is established | Separate gene validity from variant pathogenicity |
Clinical meaning emerges where genotype, phenotype, population and utility intersect.
A molecular finding becomes decision-ready only when independent evidence dimensions converge and the remaining uncertainty is visible.
READINESS
Weak phenotype turns precise sequence into imprecise interpretation.
Represent observable features
Use specific terms, onset, severity, progression, absent findings and measurement method rather than broad labels.
Preserve disease trajectory
Phenotype changes with age, treatment and environment; a snapshot can hide the pattern that makes a variant relevant.
Link relationships and uncertainty
Record affected status, ages, testing, reproductive history, ancestry and limits of available information.
| Phenotype element | High-value representation | Failure mode | Interpretive consequence |
|---|---|---|---|
| Feature | Specific observable abnormality with method | Umbrella diagnosis copied without verification | False gene prioritization |
| Onset | Age or interval with confidence | “Congenital” or “adult” used imprecisely | Incorrect inheritance or penetrance model |
| Absent feature | Explicitly assessed and not present | Not mentioned treated as absent | Overconfident exclusion |
| Trajectory | Stable, episodic or progressive course | Single encounter treated as full phenotype | Weak disease matching |
| Treatment effect | Exposure, adherence, response and time | Response separated from context | Misleading genotype–response claim |
Automation can rank evidence. It cannot erase the need for an accountable clinical judgment.
A technically correct result can still be clinically useless.
Resolve or refine a diagnosis
End the diagnostic odyssey, avoid inappropriate testing and clarify surveillance or specialist care.
Change treatment choice
Select, avoid or dose an intervention when evidence supports a genotype-informed decision.
Inform relatives and reproduction
Support cascade testing, risk clarification and reproductive decisions with consent-sensitive communication.
Create value beyond intervention
Provide explanation, planning or identity value while distinguishing it from proven clinical benefit.
Before testing, define what decision could change for each plausible result—including negative, uncertain and incidental findings—and who owns follow-through.
Genotype-informed prescribing requires a complete chain from allele to medication action.
| Layer | Question | Required control | Failure mode |
|---|---|---|---|
| Genotype | Were relevant alleles and variant classes measured? | Platform coverage, copy number and structural variant capability | Untested allele interpreted as normal |
| Diplotype / phenotype | How do alleles translate into predicted function? | Versioned translation rules and ambiguity handling | Phasing or novel allele uncertainty hidden |
| Drug evidence | Does predicted function alter exposure, response or risk? | Drug-specific evidence, population and treatment context | Biological association treated as dosing guidance |
| Clinical context | Do comedications, organ function or disease change interpretation? | Medication reconciliation and current patient state | Genotype overrides stronger clinical variables |
| Action | What decision follows and how strong is the recommendation? | Current guidance, alternatives, monitoring and prescriber ownership | Static report without workflow integration |
A risk score is a model estimate, not a diagnosis and not a universal biological constant.
Variants, weights and outcome
Document the training population, phenotype definition, statistical method and intended use.
Performance across populations
Calibration and discrimination may shift with ancestry, environment, healthcare access and baseline risk.
Combine score with context
Age, sex, family history, clinical factors and time horizon shape the probability relevant to care.
Test decision thresholds
Clinical value depends on whether score-guided action improves outcomes without disproportionate harm or burden.
Genomic data is personal, familial, persistent and difficult to make anonymous.
Choice must match real data use
Explain test scope, secondary findings, future use, sharing, retention, recontact and practical withdrawal limits.
One result can implicate others
Balance confidentiality, communication, relational duties and local legal or professional requirements.
Protect the full data lifecycle
Control access, linkage, export, cloud processing, audit, retention and incident response—not only raw sequence.
Measure who benefits and who is missed
Reference diversity, test access, language, counseling, follow-up and affordability affect precision in practice.
The same molecular result can support a different decision in a different person.
Rare disease diagnosis
Phenotype depth, family structure, inheritance, variant class and reanalysis determine whether sequencing ends or extends the diagnostic journey.
Somatic tumor profiling
Tumor content, assay scope, clonal structure, germline possibility, treatment evidence and disease stage shape actionability.
Pharmacogenomic prescribing
Genotype, functional translation, comedications, organ function and current guidance must reach the prescriber workflow.
Hereditary risk assessment
Variant evidence, penetrance, family history, surveillance options and patient preferences determine clinical value.
Newborn genomic screening
Population benefit requires high-confidence conditions, timely intervention, confirmatory pathways and long-term family support.
Polygenic risk stratification
Model transportability, absolute risk, threshold, intervention and equity must be validated before broad clinical use.
The genome may be stable. Its interpretation is not.
| Trigger | What changed | Review action | Communication control |
|---|---|---|---|
| New gene–disease evidence | Previously uncertain gene relationship becomes stronger or weaker | Reprioritize relevant variants and reassess fit | Define recontact policy and clinical significance threshold |
| Variant reclassification | Evidence changes classification or confidence | Review prior report, current phenotype and management effect | Notify accountable clinician through traceable workflow |
| Phenotype evolution | New features, age or disease course alter the match | Update phenotype and rerun prioritization | Explain why the same data now supports a different conclusion |
| Pipeline improvement | New calling method detects a previously missed variant class | Reprocess where analytically justified | Distinguish reinterpretation from new laboratory testing |
| New intervention | Finding becomes clinically actionable | Reassess utility, eligibility and current patient context | Assign follow-up and shared decision-making owner |
Precision fails when a correct signal crosses an uncontrolled interface.
| Interface | Failure | Early signal | Control |
|---|---|---|---|
| Person ↔ sample | Identity, consent or specimen history is uncertain | Metadata mismatch or unexplained sample quality | Positive identification and end-to-end provenance |
| Sequence ↔ variant | Technical artifact becomes a molecular claim | Low-quality region, imbalance or pipeline disagreement | Quality thresholds, orthogonal confirmation and limitations |
| Variant ↔ disease | Association is mistaken for causation | Weak gene validity or phenotype mismatch | Structured evidence and alternative hypothesis review |
| Result ↔ decision | Report is accurate but not actionable | No owner, pathway or defined response | Pre-test decision mapping and closed-loop follow-up |
| Model ↔ population | Performance degrades in underrepresented groups | Calibration drift or unequal uncertain-result rates | Stratified validation, monitoring and corrective access design |
Genomics and precision medicine, precisely bounded.
What is genomics?
Genomics studies genomes, including sequence, variation, structure, function and interactions. It can be applied to individuals, families, tumors, microbes or populations.
What is precision medicine?
Precision medicine integrates molecular, clinical, environmental and personal information to improve prevention, diagnosis or treatment decisions for a defined person or group.
Does a pathogenic variant always cause disease?
No. Disease expression can depend on penetrance, inheritance, age, other variants, environment and clinical context. Classification and individual risk are related but distinct judgments.
What is a variant of uncertain significance?
It is a variant for which available evidence is insufficient or conflicting for a confident benign or pathogenic classification. It generally should not be treated as a definitive molecular diagnosis.
Why can genomic interpretations change?
Gene–disease knowledge, population data, functional evidence, phenotype and classification methods evolve. Reinterpretation should use a defined governance and recontact process.
Is this page medical or genetic counseling advice?
No. It is an industry and evidence-system model. Individual testing, diagnosis, treatment and familial decisions require qualified professionals and current local guidance.