eduKateSG · Why Science?
Collect genomic samples fairly, preserve them well and keep consent, privacy and future science aligned
Read FDA’s final March 2018 ICH E18 guidance through unbiased sampling, timing, specimen quality, consent, coding, storage, genomic data standards, analysis planning, transparency and international studies.
Genomic discovery often depends on questions nobody could fully predict when a trial began. A well-planned sample may later explain why people respond differently, while a biased or poorly stored collection may make a promising question impossible to answer.
FDA’s final March 2018 ICH E18 guidance provides harmonised principles for genomic sampling and management of genomic data in clinical studies. It joins science with informed consent, privacy, data protection and transparency.
Did you know? Collecting samples from only a convenient subset can create bias before any sequencing begins. The decision about whom to invite, when to sample and how long to store material is part of the study design.
For students, E18 shows that good data begin before the laboratory. Representation, consent, specimen handling and metadata shape every later conclusion.
Chapter index
Chapters 1–9: March 2018 final guidance, unbiased collection, sampling time, specimen quality and representativeness
- 1. E18 encourages genomic research within clinical studies
- 2. The March 2018 guidance is final
- 3. Sampling objectives should be written early
- 4. Unbiased collection protects representativeness
- 5. Timing changes biological interpretation
- 6. Specimen type must match the question
- 7. Preanalytical quality begins at collection
- 8. Consent should match intended and future use
- 9. Privacy protection needs a traceability design
Chapters 10–17: consent, privacy, coding, storage, genomic data, analysis, transparency and reuse
- 10. Withdrawal rules depend on the sample state
- 11. Storage plans preserve scientific usability
- 12. Genomic data need durable metadata
- 13. Analysis plans should anticipate population structure
- 14. Global studies face local differences
- 15. Return of findings needs a defined pathway
- 16. Transparency helps participants and science
- 17. A complete E18 receipt preserves the chain
CHAPTER 1 OF 17 · Plan the sample
1. E18 encourages genomic research within clinical studies
The guidance supports consistent collection and management of DNA- and RNA-related samples and data so genomic questions can be studied across global development programmes. It follows the genomic concepts defined in ICH E15.
The practical value of e18 encourages genomic research within clinical studies appears when the scientific claim and its boundary stay together. The guidance supports consistent collection and management of DNA- and RNA-related samples and data so genomic questions can be studied across global development programmes. It follows the genomic concepts defined in ICH E15. Keep this boundary visible: A larger repository cannot correct biased enrolment or inconsistent sampling.
Try this
Build an evidence map for e18 encourages genomic research within clinical studies. Compare invited and sampled participants for e18 encourages genomic research within clinical studies and explain every imbalance. Add the owner and decision it supports. Keep this limit: A larger repository cannot correct biased enrolment or inconsistent sampling.
A strong science note separates observation from interpretation. Compare invited and sampled participants for e18 encourages genomic research within clinical studies and explain every imbalance. Record the source, version or date, population or device context, method, result and one uncertainty. If the official record does not show a field, write “not shown” instead of filling the gap from memory.
Evidence checkpoint
Compare invited and sampled participants for e18 encourages genomic research within clinical studies and explain every imbalance. Keep this boundary visible: A larger repository cannot correct biased enrolment or inconsistent sampling.
FDA lists E18 as final guidance dated March 2018 under docket FDA-2016-D-1255, issued by CDER, CDRH and CBER. Its main objective is harmonised, unbiased sampling and optimal use of genomic samples and data.
In the march 2018 guidance is final, evidence has to match the decision being proposed. FDA lists E18 as final guidance dated March 2018 under docket FDA-2016-D-1255, issued by CDER, CDRH and CBER. Its main objective is harmonised, unbiased sampling and optimal use of genomic samples and data. Preserve this guardrail: Genomic data remain sensitive even when obvious identifiers are removed.
Try this
Test the reasoning for the march 2018 guidance is final. Draw the consent-to-repository workflow for the march 2018 guidance is final, including withdrawal limits. Write the strongest alternative explanation. Respect this boundary: Genomic data remain sensitive even when obvious identifiers are removed.
A strong science note separates observation from interpretation. Draw the consent-to-repository workflow for the march 2018 guidance is final, including withdrawal limits. Record the source, version or date, population or device context, method, result and one uncertainty. If the official record does not show a field, write “not shown” instead of filling the gap from memory.
Evidence checkpoint
Draw the consent-to-repository workflow for the march 2018 guidance is final, including withdrawal limits. Keep this boundary visible: Genomic data remain sensitive even when obvious identifiers are removed.
A protocol should explain why genomic samples are collected, which analyses are planned or possible and how results could inform drug response, disease biology or future research. Early planning aligns volume, handling, consent and resources.
A trustworthy explanation of sampling objectives should be written early shows both what the evidence supports and what remains open. A protocol should explain why genomic samples are collected, which analyses are planned or possible and how results could inform drug response, disease biology or future research. Early planning aligns volume, handling, consent and resources. Remember: Future use must stay within consent, law and repository governance.
Try this
Audit the record for sampling objectives should be written early. Audit preanalytical handling for sampling objectives should be written early from collection through long-term storage. Mark every assumption and missing link. Preserve this caution: Future use must stay within consent, law and repository governance.
A strong science note separates observation from interpretation. Audit preanalytical handling for sampling objectives should be written early from collection through long-term storage. Record the source, version or date, population or device context, method, result and one uncertainty. If the official record does not show a field, write “not shown” instead of filling the gap from memory.
Evidence checkpoint
Audit preanalytical handling for sampling objectives should be written early from collection through long-term storage. Keep this boundary visible: Future use must stay within consent, law and repository governance.
Inviting all eligible participants or using a justified sampling strategy reduces selection bias. Sites, regions or participant groups with lower collection rates can distort allele frequencies and biomarker relationships.
The practical value of unbiased collection protects representativeness appears when the scientific claim and its boundary stay together. Inviting all eligible participants or using a justified sampling strategy reduces selection bias. Sites, regions or participant groups with lower collection rates can distort allele frequencies and biomarker relationships. Keep this boundary visible: A stored sample is valuable only when integrity and metadata remain adequate.
Try this
Build an evidence map for unbiased collection protects representativeness. Reproduce one genomic result for unbiased collection protects representativeness from raw data, metadata and pipeline versions. Add the owner and decision it supports. Keep this limit: A stored sample is valuable only when integrity and metadata remain adequate.
A strong science note separates observation from interpretation. Reproduce one genomic result for unbiased collection protects representativeness from raw data, metadata and pipeline versions. Record the source, version or date, population or device context, method, result and one uncertainty. If the official record does not show a field, write “not shown” instead of filling the gap from memory.
Evidence checkpoint
Reproduce one genomic result for unbiased collection protects representativeness from raw data, metadata and pipeline versions. Keep this boundary visible: A stored sample is valuable only when integrity and metadata remain adequate.
CHAPTER 5 OF 17 · Protect the participant
5. Timing changes biological interpretation
Baseline, on-treatment, progression and recovery samples answer different questions, especially for RNA or tumour material. Timing relative to dosing, meals, procedures and disease events should be recorded.
In timing changes biological interpretation, evidence has to match the decision being proposed. Baseline, on-treatment, progression and recovery samples answer different questions, especially for RNA or tumour material. Timing relative to dosing, meals, procedures and disease events should be recorded. Preserve this guardrail: A larger repository cannot correct biased enrolment or inconsistent sampling.
Try this
Test the reasoning for timing changes biological interpretation. Compare invited and sampled participants for timing changes biological interpretation and explain every imbalance. Write the strongest alternative explanation. Respect this boundary: A larger repository cannot correct biased enrolment or inconsistent sampling.
A strong science note separates observation from interpretation. Compare invited and sampled participants for timing changes biological interpretation and explain every imbalance. Record the source, version or date, population or device context, method, result and one uncertainty. If the official record does not show a field, write “not shown” instead of filling the gap from memory.
Evidence checkpoint
Compare invited and sampled participants for timing changes biological interpretation and explain every imbalance. Keep this boundary visible: A larger repository cannot correct biased enrolment or inconsistent sampling.
Blood, saliva, tissue and other materials differ in cellular source, DNA or RNA yield and invasiveness. Germline and somatic questions may require different specimens and handling.
A trustworthy explanation of specimen type must match the question shows both what the evidence supports and what remains open. Blood, saliva, tissue and other materials differ in cellular source, DNA or RNA yield and invasiveness. Germline and somatic questions may require different specimens and handling. Remember: Genomic data remain sensitive even when obvious identifiers are removed.
Try this
Audit the record for specimen type must match the question. Draw the consent-to-repository workflow for specimen type must match the question, including withdrawal limits. Mark every assumption and missing link. Preserve this caution: Genomic data remain sensitive even when obvious identifiers are removed.
A strong science note separates observation from interpretation. Draw the consent-to-repository workflow for specimen type must match the question, including withdrawal limits. Record the source, version or date, population or device context, method, result and one uncertainty. If the official record does not show a field, write “not shown” instead of filling the gap from memory.
Evidence checkpoint
Draw the consent-to-repository workflow for specimen type must match the question, including withdrawal limits. Keep this boundary visible: Genomic data remain sensitive even when obvious identifiers are removed.
CHAPTER 7 OF 17 · Protect the participant
7. Preanalytical quality begins at collection
Tube type, processing delay, temperature, stabilisation, shipment, freeze-thaw cycles and storage can change sample integrity. Standard procedures and deviation records keep laboratory variation from masquerading as biology.
The practical value of preanalytical quality begins at collection appears when the scientific claim and its boundary stay together. Tube type, processing delay, temperature, stabilisation, shipment, freeze-thaw cycles and storage can change sample integrity. Standard procedures and deviation records keep laboratory variation from masquerading as biology. Keep this boundary visible: Future use must stay within consent, law and repository governance.
Try this
Build an evidence map for preanalytical quality begins at collection. Audit preanalytical handling for preanalytical quality begins at collection from collection through long-term storage. Add the owner and decision it supports. Keep this limit: Future use must stay within consent, law and repository governance.
A strong science note separates observation from interpretation. Audit preanalytical handling for preanalytical quality begins at collection from collection through long-term storage. Record the source, version or date, population or device context, method, result and one uncertainty. If the official record does not show a field, write “not shown” instead of filling the gap from memory.
Evidence checkpoint
Audit preanalytical handling for preanalytical quality begins at collection from collection through long-term storage. Keep this boundary visible: Future use must stay within consent, law and repository governance.
CHAPTER 8 OF 17 · Protect the participant
8. Consent should match intended and future use
Suitable informed consent should address genomic research, storage duration, possible future analyses, data sharing, withdrawal limits and relevant return-of-results arrangements under local requirements. The operational system must be able to honour what is promised.
In consent should match intended and future use, evidence has to match the decision being proposed. Suitable informed consent should address genomic research, storage duration, possible future analyses, data sharing, withdrawal limits and relevant return-of-results arrangements under local requirements. The operational system must be able to honour what is promised. Preserve this guardrail: A stored sample is valuable only when integrity and metadata remain adequate.
Try this
Test the reasoning for consent should match intended and future use. Reproduce one genomic result for consent should match intended and future use from raw data, metadata and pipeline versions. Write the strongest alternative explanation. Respect this boundary: A stored sample is valuable only when integrity and metadata remain adequate.
A strong science note separates observation from interpretation. Reproduce one genomic result for consent should match intended and future use from raw data, metadata and pipeline versions. Record the source, version or date, population or device context, method, result and one uncertainty. If the official record does not show a field, write “not shown” instead of filling the gap from memory.
Evidence checkpoint
Reproduce one genomic result for consent should match intended and future use from raw data, metadata and pipeline versions. Keep this boundary visible: A stored sample is valuable only when integrity and metadata remain adequate.
Coding reduces direct identification while allowing controlled linkage where consent and law permit. Key custody, access control and re-identification risk should reflect the sensitivity and future use of genomic information.
A trustworthy explanation of privacy protection needs a traceability design shows both what the evidence supports and what remains open. Coding reduces direct identification while allowing controlled linkage where consent and law permit. Key custody, access control and re-identification risk should reflect the sensitivity and future use of genomic information. Remember: A larger repository cannot correct biased enrolment or inconsistent sampling.
Try this
Audit the record for privacy protection needs a traceability design. Compare invited and sampled participants for privacy protection needs a traceability design and explain every imbalance. Mark every assumption and missing link. Preserve this caution: A larger repository cannot correct biased enrolment or inconsistent sampling.
A strong science note separates observation from interpretation. Compare invited and sampled participants for privacy protection needs a traceability design and explain every imbalance. Record the source, version or date, population or device context, method, result and one uncertainty. If the official record does not show a field, write “not shown” instead of filling the gap from memory.
Evidence checkpoint
Compare invited and sampled participants for privacy protection needs a traceability design and explain every imbalance. Keep this boundary visible: A larger repository cannot correct biased enrolment or inconsistent sampling.
A participant may be able to stop future use of retained coded material, but analyses already completed or irreversibly anonymised data may not be retractable. These limits should be explained before participation and recorded consistently.
The practical value of withdrawal rules depend on the sample state appears when the scientific claim and its boundary stay together. A participant may be able to stop future use of retained coded material, but analyses already completed or irreversibly anonymised data may not be retractable. These limits should be explained before participation and recorded consistently. Keep this boundary visible: Genomic data remain sensitive even when obvious identifiers are removed.
Try this
Build an evidence map for withdrawal rules depend on the sample state. Draw the consent-to-repository workflow for withdrawal rules depend on the sample state, including withdrawal limits. Add the owner and decision it supports. Keep this limit: Genomic data remain sensitive even when obvious identifiers are removed.
A strong science note separates observation from interpretation. Draw the consent-to-repository workflow for withdrawal rules depend on the sample state, including withdrawal limits. Record the source, version or date, population or device context, method, result and one uncertainty. If the official record does not show a field, write “not shown” instead of filling the gap from memory.
Evidence checkpoint
Draw the consent-to-repository workflow for withdrawal rules depend on the sample state, including withdrawal limits. Keep this boundary visible: Genomic data remain sensitive even when obvious identifiers are removed.
Repositories need inventory, environmental monitoring, chain of custody, disaster planning, quality checks and rules for depletion. Long retention can enable future science only if samples and metadata remain interpretable.
In storage plans preserve scientific usability, evidence has to match the decision being proposed. Repositories need inventory, environmental monitoring, chain of custody, disaster planning, quality checks and rules for depletion. Long retention can enable future science only if samples and metadata remain interpretable. Preserve this guardrail: Future use must stay within consent, law and repository governance.
Try this
Test the reasoning for storage plans preserve scientific usability. Audit preanalytical handling for storage plans preserve scientific usability from collection through long-term storage. Write the strongest alternative explanation. Respect this boundary: Future use must stay within consent, law and repository governance.
A strong science note separates observation from interpretation. Audit preanalytical handling for storage plans preserve scientific usability from collection through long-term storage. Record the source, version or date, population or device context, method, result and one uncertainty. If the official record does not show a field, write “not shown” instead of filling the gap from memory.
Evidence checkpoint
Audit preanalytical handling for storage plans preserve scientific usability from collection through long-term storage. Keep this boundary visible: Future use must stay within consent, law and repository governance.
Reference genome, build, assay platform, library preparation, software, pipeline version, quality thresholds and phenotype definitions affect meaning. Data without this context may be impossible to reproduce or combine.
A trustworthy explanation of genomic data need durable metadata shows both what the evidence supports and what remains open. Reference genome, build, assay platform, library preparation, software, pipeline version, quality thresholds and phenotype definitions affect meaning. Data without this context may be impossible to reproduce or combine. Remember: A stored sample is valuable only when integrity and metadata remain adequate.
Try this
Audit the record for genomic data need durable metadata. Reproduce one genomic result for genomic data need durable metadata from raw data, metadata and pipeline versions. Mark every assumption and missing link. Preserve this caution: A stored sample is valuable only when integrity and metadata remain adequate.
A strong science note separates observation from interpretation. Reproduce one genomic result for genomic data need durable metadata from raw data, metadata and pipeline versions. Record the source, version or date, population or device context, method, result and one uncertainty. If the official record does not show a field, write “not shown” instead of filling the gap from memory.
Evidence checkpoint
Reproduce one genomic result for genomic data need durable metadata from raw data, metadata and pipeline versions. Keep this boundary visible: A stored sample is valuable only when integrity and metadata remain adequate.
CHAPTER 13 OF 17 · Preserve future value
13. Analysis plans should anticipate population structure
Ancestry, relatedness, batch effects, multiplicity, rare variants and missingness can bias associations. Statistical methods and sensitivity analyses should be chosen for the question and the composition of the sampled population.
The practical value of analysis plans should anticipate population structure appears when the scientific claim and its boundary stay together. Ancestry, relatedness, batch effects, multiplicity, rare variants and missingness can bias associations. Statistical methods and sensitivity analyses should be chosen for the question and the composition of the sampled population. Keep this boundary visible: A larger repository cannot correct biased enrolment or inconsistent sampling.
Try this
Build an evidence map for analysis plans should anticipate population structure. Compare invited and sampled participants for analysis plans should anticipate population structure and explain every imbalance. Add the owner and decision it supports. Keep this limit: A larger repository cannot correct biased enrolment or inconsistent sampling.
A strong science note separates observation from interpretation. Compare invited and sampled participants for analysis plans should anticipate population structure and explain every imbalance. Record the source, version or date, population or device context, method, result and one uncertainty. If the official record does not show a field, write “not shown” instead of filling the gap from memory.
Evidence checkpoint
Compare invited and sampled participants for analysis plans should anticipate population structure and explain every imbalance. Keep this boundary visible: A larger repository cannot correct biased enrolment or inconsistent sampling.
Consent language, export rules, privacy law, ethics review and cultural expectations can vary across regions. Harmonised scientific principles still require region-specific implementation and documented limits on transfer or reuse.
In global studies face local differences, evidence has to match the decision being proposed. Consent language, export rules, privacy law, ethics review and cultural expectations can vary across regions. Harmonised scientific principles still require region-specific implementation and documented limits on transfer or reuse. Preserve this guardrail: Genomic data remain sensitive even when obvious identifiers are removed.
Try this
Test the reasoning for global studies face local differences. Draw the consent-to-repository workflow for global studies face local differences, including withdrawal limits. Write the strongest alternative explanation. Respect this boundary: Genomic data remain sensitive even when obvious identifiers are removed.
A strong science note separates observation from interpretation. Draw the consent-to-repository workflow for global studies face local differences, including withdrawal limits. Record the source, version or date, population or device context, method, result and one uncertainty. If the official record does not show a field, write “not shown” instead of filling the gap from memory.
Evidence checkpoint
Draw the consent-to-repository workflow for global studies face local differences, including withdrawal limits. Keep this boundary visible: Genomic data remain sensitive even when obvious identifiers are removed.
CHAPTER 15 OF 17 · Preserve future value
15. Return of findings needs a defined pathway
Plans should distinguish research results, analytically valid findings and clinically actionable information, and identify who can interpret and communicate them. A research assay result may not be suitable for clinical decisions.
A trustworthy explanation of return of findings needs a defined pathway shows both what the evidence supports and what remains open. Plans should distinguish research results, analytically valid findings and clinically actionable information, and identify who can interpret and communicate them. A research assay result may not be suitable for clinical decisions. Remember: Future use must stay within consent, law and repository governance.
Try this
Audit the record for return of findings needs a defined pathway. Audit preanalytical handling for return of findings needs a defined pathway from collection through long-term storage. Mark every assumption and missing link. Preserve this caution: Future use must stay within consent, law and repository governance.
A strong science note separates observation from interpretation. Audit preanalytical handling for return of findings needs a defined pathway from collection through long-term storage. Record the source, version or date, population or device context, method, result and one uncertainty. If the official record does not show a field, write “not shown” instead of filling the gap from memory.
Evidence checkpoint
Audit preanalytical handling for return of findings needs a defined pathway from collection through long-term storage. Keep this boundary visible: Future use must stay within consent, law and repository governance.
CHAPTER 16 OF 17 · Preserve future value
16. Transparency helps participants and science
Clear protocols, consent materials, repository governance and reporting of genomic methods support trust. Negative or inconclusive findings can still be valuable when sampling quality and limitations are disclosed.
The practical value of transparency helps participants and science appears when the scientific claim and its boundary stay together. Clear protocols, consent materials, repository governance and reporting of genomic methods support trust. Negative or inconclusive findings can still be valuable when sampling quality and limitations are disclosed. Keep this boundary visible: A stored sample is valuable only when integrity and metadata remain adequate.
Try this
Build an evidence map for transparency helps participants and science. Reproduce one genomic result for transparency helps participants and science from raw data, metadata and pipeline versions. Add the owner and decision it supports. Keep this limit: A stored sample is valuable only when integrity and metadata remain adequate.
A strong science note separates observation from interpretation. Reproduce one genomic result for transparency helps participants and science from raw data, metadata and pipeline versions. Record the source, version or date, population or device context, method, result and one uncertainty. If the official record does not show a field, write “not shown” instead of filling the gap from memory.
Evidence checkpoint
Reproduce one genomic result for transparency helps participants and science from raw data, metadata and pipeline versions. Keep this boundary visible: A stored sample is valuable only when integrity and metadata remain adequate.
CHAPTER 17 OF 17 · Preserve future value
17. A complete E18 receipt preserves the chain
The receipt records purpose, eligibility, consent version, collection rate, specimen and timepoint, handling deviations, coding, storage, withdrawals, assay and pipeline metadata, analysis population, sharing, return-of-results policy and final disposition. It connects the participant to a defensible scientific record without unnecessary exposure.
In a complete e18 receipt preserves the chain, evidence has to match the decision being proposed. The receipt records purpose, eligibility, consent version, collection rate, specimen and timepoint, handling deviations, coding, storage, withdrawals, assay and pipeline metadata, analysis population, sharing, return-of-results policy and final disposition. It connects the participant to a defensible scientific record without unnecessary exposure. Preserve this guardrail: A larger repository cannot correct biased enrolment or inconsistent sampling.
Try this
Test the reasoning for a complete e18 receipt preserves the chain. Compare invited and sampled participants for a complete e18 receipt preserves the chain and explain every imbalance. Write the strongest alternative explanation. Respect this boundary: A larger repository cannot correct biased enrolment or inconsistent sampling.
A strong science note separates observation from interpretation. Compare invited and sampled participants for a complete e18 receipt preserves the chain and explain every imbalance. Record the source, version or date, population or device context, method, result and one uncertainty. If the official record does not show a field, write “not shown” instead of filling the gap from memory.
Evidence checkpoint
Compare invited and sampled participants for a complete e18 receipt preserves the chain and explain every imbalance. Keep this boundary visible: A larger repository cannot correct biased enrolment or inconsistent sampling.
E18 genomic sample lifecycle
| Question | Evidence to capture | Boundary to retain |
|---|---|---|
| Who is sampled? | Eligibility, invitation and collection rates | Convenience can create bias |
| What is collected? | Specimen, timing and preanalytical controls | Handling can change biology |
| How is it protected? | Consent, coding, privacy and withdrawal | Promises must be operational |
| Can it be reused? | Storage, metadata, pipelines and governance | Future value needs durable context |
Science learning becomes useful when a student can show where a claim came from, why it applies and where it stops. Continue with the eduKate Science Learning Hub, compare choices in the Education Hub, and connect these habits to career pathways built from subject capabilities.
E18 makes genomic research stronger by treating every sample as both scientific material and a participant-linked responsibility. Thoughtful collection, careful custody and transparent data management create evidence that can travel into future questions without losing its integrity.
Primary sources and verification date
Source status checked 26 October 2026. FDA guidance describes the agency’s current thinking and does not replace applicable statutes, regulations or product-specific communication with FDA.
- FDA: ICH E18 genomic sampling and data management — Final guidance, March 2018; docket FDA-2016-D-1255.
- FDA: Full ICH E18 guidance PDF — Official sampling, consent, privacy and genomic-data-management principles.
- FDA: ICH E15 pharmacogenomics definitions — Definitions and coding concepts underpinning E18.