Richtlinien zur Einreichung von Proben
Build one evidence chain from target perturbation to molecular and organoid-level consequences.
A Candidate Becomes More Credible When Three Evidence Layers Agree
The strongest study does not ask whether one measurement changed. It asks whether the intervention occurred, whether the expected molecular program moved, and whether a predefined organoid phenotype changed in the same experimental context. Agreement across these layers reduces the chance that a culture artifact, clone effect, or delivery response is mistaken for target biology.
| Evidence layer | Question answered | Typical readout | What it does not prove alone |
|---|---|---|---|
| Intervention verification | Did the intended perturbation occur in the tested material? | Targeted amplicon or Sanger sequencing for a genomic edit; target-expression check for knockdown, activation, or overexpression | That the observed phenotype is caused only by the target |
| Transcriptomic consequence | Which genes, pathways, or cell programs changed after perturbation? | Matched bulk RNA-seq as the default; single-cell RNA-seq when cell-state resolution is essential | Direct binding, direct regulation, or a unique causal route |
| Organoid phenotype | Did the model-level feature specified before the study change? | Morphology, growth, differentiation, marker localization, barrier or other agreed in vitro measurement | General biological relevance beyond the tested model and conditions |
The three-layer model also makes discordance informative. A verified edit with no reproducible RNA response may indicate that the target is buffered in the selected model, that the sampling time missed the response, or that the measured phenotype sits outside the affected pathway. A strong transcriptional shift without intervention verification leaves a simpler technical explanation unresolved. The report keeps these alternatives visible instead of forcing every result into a positive narrative.
Define the Validation Claim Before Choosing the Perturbation
The same gene can support different hypotheses, and each hypothesis requires a different contrast. A loss-of-function study asks whether reducing target activity removes a program or phenotype. A gain-of-function study asks whether added or increased activity is sufficient to produce one. A rescue study asks whether restoring the target reverses a change observed in a perturbed background. These are related experiments, but they are not interchangeable.
Project scoping records the candidate source, organoid model, background genotype, expected direction of effect, relevant time window, primary phenotype, and decision the data must support. It also defines what would count as a negative or inconclusive result. This prevents a large post hoc analysis from substituting for an answerable biological question.
Where an existing model is supplied, we review its construction and validation history before accepting the perturbation label as fact. Where model engineering is needed, the route is selected around the hypothesis and organoid biology. Knockdown, knockout, transcriptional repression or activation, and overexpression may produce different strengths, durations, compensatory responses, and selection pressures. Availability is confirmed during feasibility review rather than presented as a universal menu.
Best for: focused validation of one or a small set of nominated genes in an organoid model with a defined comparison and measurable research phenotype. Not for: unbiased target discovery across a large library, projects that have no suitable control model, or requests to infer causality from a single expression snapshot. A dedicated screen or staged discovery workflow is more appropriate when the candidate list is still broad.
Choose a Perturbation Route That Fits the Biological Question
Perturbation design is a source of evidence, not merely sample preparation. The route should create the biological state needed by the hypothesis while preserving enough model quality for downstream comparison. We therefore evaluate whether the project needs a stable or transient change, a mixed population or isolated clone, and complete loss, partial reduction, activation, or restoration.
| Perturbation route | Useful research question | Main design advantage | Important interpretation risk |
|---|---|---|---|
| Knockdown or transcriptional repression | Does partial target reduction alter the selected program? | Can model reduced activity when complete loss is not viable | Residual expression and reagent-specific effects can weaken the conclusion |
| Targeted knockout | Is the gene required for the measured state or phenotype? | Creates a direct loss-of-function contrast when edited material remains viable | Clonal selection, compensatory adaptation, or mixed alleles can shape the result |
| Activation or overexpression | Is increased target activity sufficient to shift the model? | Tests directionality from the opposite side of a loss-of-function hypothesis | Non-physiologic expression can create effects unrelated to endogenous regulation |
| Rescue or complementation | Does restoring target activity reverse the perturbed state? | Adds a strong orthogonal test of target dependence | Expression level, construct design, and timing must be comparable to the original hypothesis |
Multiple independent perturbation reagents or independently derived edited populations are valuable when feasible because concordant effects are harder to attribute to one reagent or one clone. However, simply increasing the number of technical replicates does not provide that protection. The design distinguishes independent biological preparations, independent engineering events, culture wells, sequencing libraries, and repeated measurements.
For pooled material, the intervention check should describe the allele or expression mixture represented in the sequenced RNA. For clonal material, more than one clone and an appropriate parental or mock-engineered control help separate target biology from clone-specific adaptation. The final plan documents which route was chosen and what limitation it introduces.
Confirm the Intervention Before Interpreting the Transcriptome
An organoid labeled "knockout" or "overexpression" is still a hypothesis until the tested batch is verified. Confirmation should match the intervention. A genomic edit may require amplification and sequencing of the target region; a knockdown or activation experiment may require target-expression measurement; an overexpression construct may require confirmation that the intended transcript is present in the relevant material.
Targeted DNA sequencing is used to answer a focused question: which alleles or sequence changes are present at the intended locus in the organoids entering the comparison? It is not automatically expanded into whole-exome or whole-genome analysis, and it does not by itself establish functional loss. When a mixed population is used, allele fractions and predicted consequences must be interpreted alongside the possibility that unedited or differently edited cells contribute to the RNA and phenotype.
Verification is planned on material that represents the profiled culture state whenever possible. A historical edit report from an earlier passage does not rule out drift, selection, or changes in population composition. Batch identity, passage, collection time, and sample relationships are therefore carried from the intervention record into the sequencing manifest.
Use orthogonal perturbations and rescue logic to strengthen target-specific interpretation.
Run One Coordinated Workflow from Model Review to Evidence Package
The service keeps engineering history, sampling, RNA sequencing, phenotype metadata, and analysis under one comparison map. Each checkpoint is designed to stop an uninterpretable study before additional organoid material is consumed.
Coordinate hypothesis, controls, intervention verification, RNA sequencing and phenotype interpretation.
- Frame the target hypothesis. Define the expected direction of effect, organoid background, primary contrast, phenotype, and evidence needed to support or reject the claim.
- Review the model and perturbation route. Confirm whether established organoids can be used or whether a model-engineering module is required. Record parental background, passage, matrix, medium, and intervention history.
- Establish matched controls. Select non-targeting, mock, parental, isogenic, rescue, or independent-reagent controls according to the intervention. Balance culture and processing batches across groups.
- Verify the intervention. Perform the agreed target-locus DNA check or target-expression check on representative material. Review mixed alleles, incomplete reduction, and unexpected sequence outcomes before profiling.
- Collect RNA and phenotype data. Harvest matched biological replicates at the selected time point or series. Capture the predefined organoid measurement without allowing it to become an unplanned endpoint search.
- Sequence and analyze the response. Apply the agreed RNA-seq route, review read and sample relationships, model the intended contrast, and connect differential genes or pathways to the phenotype record.
- Integrate and report the evidence. Classify findings as concordant, discordant, negative, or limited; distinguish direct observations from model-derived results and follow-up hypotheses.
If intervention verification fails, the study does not move forward as though the comparison were valid. The route may be revised, material may be re-established, or the project may be reported as a feasibility outcome. That decision protects the customer from paying for an RNA-seq dataset that cannot answer the original question.
Match Samples and Controls to the Perturbation, Not a Generic Checklist
Sample requirements depend on organoid type, intervention, RNA-seq design, and whether DNA confirmation or phenotyping is included. We confirm quantities after feasibility review; public supplier numbers and conditions are not transferred into a CD Genomics promise.
| Material or record | What to provide or define | QC and matching focus | Why it matters |
|---|---|---|---|
| Established organoid model | Source type, tissue or differentiation route, passage, matrix, medium, culture history, and available characterization | Identity, culture status, contamination record, and suitability for the requested intervention | A poorly matched or unstable model can obscure a genuine target effect |
| Perturbed and control organoids | Intervention construct or reagent, selection history, clone or pool designation, collection plan, and relationship to parental material | Comparable passage, batch, handling, and predefined replicate structure | The comparison must isolate the target variable rather than a culture-history difference |
| DNA or cell material for edit confirmation | Representative material from the profiled batch and the target-region sequence or edit design | Amplifiability, locus specificity, allele mixture, and sample identity | Links the RNA and phenotype to the edit actually present at collection |
| RNA or harvest-ready organoids | Matched groups collected under a shared clock and documented preservation route | RNA quality, yield, contamination, sample swaps, and library compatibility | Prevents handling time or RNA degradation from becoming the dominant contrast |
| Phenotype data or assay plan | Pre-specified endpoint, acquisition method, units, normalization, blinding or analysis rule where relevant | Same culture window and sample map as molecular profiling | Allows molecular changes to be tested against an interpretable model-level outcome |
| Existing sequencing data | Raw or processed files, sample sheet, reference build, annotation, contrast definition, and QC documentation | Completeness, comparability, and ability to reconstruct the original model | Supports analysis-only work without overstating what absent experimental controls can establish |
Organoid variability should be addressed at the design level. Pooling several organoids can provide material and average local variation, but it does not create independent biological replicates. Similarly, multiple sections or images from one organoid do not replace independently prepared cultures. The manifest makes these relationships explicit so statistical confidence is not overstated.
Use RNA Sequencing to Test the Consequence, Not Just Reconfirm the Target
The core RNA-seq analysis asks how the perturbed organoid state differs from its matched control after accounting for the design. Target expression is one checkpoint, but the main value lies in evaluating downstream programs, pathway coherence, and whether the response aligns with the phenotype specified before sequencing.
Bulk RNA-seq is the default when the question concerns an average response across the organoid and the groups are compositionally comparable. The model can include donor, clone, batch, passage, or paired structure when the design supports those terms. Sample-level exploration, expression quantification, differential analysis, gene-set or pathway interpretation, and prioritized result tables are then organized around the target hypothesis rather than every mathematically possible contrast.
Single-cell RNA-seq is considered when cell-state composition, rare populations, lineage-specific effects, or mosaic perturbation would make the sample average misleading. It can show whether an apparent bulk change reflects altered expression within a cell population or a change in the abundance of that population. It also requires a stronger sampling and replication plan; thousands of cells from one preparation do not become thousands of biological replicates.
The annotation source, reference build, filtering decisions, contrast formula, and pathway resources are recorded with the deliverable. This makes the analysis reviewable and helps a later follow-up distinguish a biological difference from a pipeline or annotation change.
Connect Molecular Response with a Predefined Organoid Phenotype
Phenotype integration works best when the endpoint is chosen before the RNA results are seen. Depending on the model and question, that endpoint may be organoid formation, size or budding, growth, differentiation, marker localization, structural organization, barrier behavior, or another agreed in vitro measurement. It should have a defined acquisition and comparison rule rather than a retrospective description of whichever image looks most different.
RNA and phenotype do not need to tell an identical story to be useful. A reproducible phenotype with a coherent pathway response strengthens the working mechanism. A phenotype without a corresponding transcriptomic program may point to timing, post-transcriptional regulation, or an endpoint that is too distant from the target. A molecular response without a phenotype can indicate compensation, insufficient observation time, or a phenotype that was not sensitive to the affected biology.
The integration step therefore produces a triangulation table, not a binary "validated" label. Each candidate conclusion is supported by the intervention check, the RNA result, the phenotype record, replicate consistency, and known limitations. Direct target binding, protein activity, or a specific biochemical mechanism requires an appropriate orthogonal assay; it is not inferred from pathway enrichment alone.
Receive an Auditable Target-Validation Evidence Package
The final package is organized for both scientific review and the next experimental decision. It preserves the chain from the target hypothesis to the observed and model-derived outputs instead of delivering disconnected figures.
| Package component | What it contains | Decision value |
|---|---|---|
| Study and sample map | Target hypothesis, organoid background, perturbation route, controls, replicate relationships, collection timing, and deviations | Shows exactly which claim the experiment can and cannot address |
| Intervention-verification results | Targeted DNA or expression evidence, sample identities, allele or expression summaries, and limitations | Confirms whether the molecular and phenotype data came from the intended perturbed state |
| RNA sequencing data and QC | Agreed sequence files, sample-level QC, expression matrices, and analysis-ready metadata | Supports archive, reanalysis, and independent review of the primary molecular readout |
| Differential and pathway analysis | Predefined contrasts, effect estimates, uncertainty, gene-set summaries, and prioritized response modules | Reveals whether the perturbation changed a coherent biological program |
| Phenotype-linked evidence table | Predefined phenotype results aligned with intervention and RNA findings for each replicate or model | Makes concordance and discordance visible rather than hiding them in separate reports |
| Interpretation and limitation report | Evidence strength, alternative explanations, non-estimable comparisons, and focused follow-up options | Helps the team decide whether to advance, redesign, or stop the hypothesis |
An analysis-only project may begin with existing DNA, RNA-seq, and phenotype files. In that situation, the report clearly separates conclusions supported by the submitted design from questions that cannot be recovered after collection. Missing controls, unclear clone relationships, or confounded batches are documented rather than repaired statistically by assumption.
Select the Smallest Study That Can Answer the Target Question
More layers are not automatically better. A focused knockout versus matched-control bulk RNA-seq study may be sufficient when the organoid is compositionally stable and the expected phenotype is clear. A single-cell design becomes useful when cell-state specificity is itself the hypothesis. A rescue experiment is valuable when the key uncertainty is whether the phenotype truly depends on the target rather than the engineering event.
| Study route | Best suited to | Evidence added | When to choose another route |
|---|---|---|---|
| Perturbation verification plus bulk RNA-seq | A defined target with an organoid-wide molecular hypothesis | Target status, average transcriptional response, and pathway coherence | Choose single-cell profiling when changing cell composition would confound the average |
| Perturbation verification plus single-cell RNA-seq | Cell-type-specific, lineage, mosaic, or rare-state effects | Cell-state-resolved response and composition changes | Choose bulk RNA-seq when the model is homogeneous enough and the question is average expression |
| Two independent reagents or edited populations | Concern about reagent- or clone-specific effects | Replication across separate interventions | Add rescue when the remaining question is reversibility or target dependence |
| Rescue or complementation design | A loss-of-function phenotype that needs orthogonal support | Reversal evidence in the perturbed background | Do not use an uncontrolled overexpression comparison as a substitute for a matched rescue |
| Large pooled screen | The candidate list is broad and discovery is still required | Relative guide abundance or single-cell perturbation readouts across many targets | Do not treat this focused service page as a promise of genome-wide screening capacity |
Teams that need broader editing-focused locus assessment can review our genome editing sequencing service. When the key output is a conventional expression comparison, the mRNA sequencing service provides the underlying transcriptome workflow. For general organoid profiling across bulk, single-cell, and spatial routes, see organoid sequencing services. A target-validation project uses these capabilities within a question-led design rather than treating them as unrelated add-ons.
Illustrative Target-Validation Results
The following examples show how a project can organize its evidence. They are conceptual illustrations, not customer data, acceptance thresholds, or promised outcomes. Actual figures depend on the organoid model, intervention, readout, and data that pass the agreed review criteria.
Illustrative example: verify the intervention before interpreting downstream RNA changes.
Intervention Verification
A locus-focused view can summarize wild-type and edited alleles across an organoid pool or clone set, while a paired expression panel can show whether target reduction or activation is present in the samples entering RNA-seq. The purpose is to connect the perturbation label to the profiled material, not to imply functional success from an edit percentage alone.
Illustrative example: interpret transcriptomic response across matched donors and controls.
Transcriptomic Consequence
A sample-relationship view, differential-response plot, and pathway summary can reveal whether biological replicates separate by intervention after donor or clone structure is considered. A coherent pathway pattern is interpreted as support for a response program; it is not evidence of direct target binding.
Illustrative example: distinguish concordant, discordant and incomplete evidence packages.
Genotype–RNA–Phenotype Triangulation
An evidence matrix can place each independent perturbation beside intervention status, RNA response, phenotype direction, and limitation flags. Concordance across independent reagents or models increases confidence. Discordant rows remain visible so the team can decide whether timing, model heterogeneity, engineering history, or the original hypothesis needs revision.
Organoid Target Validation FAQs
Can you start with organoids that our laboratory has already edited?
Yes, subject to feasibility review. Provide the parental relationship, editing or expression records, clone or pool history, passage, culture conditions, and any existing DNA, RNA, or phenotype data. We will identify which verification steps are still needed to connect the supplied model to the planned comparison.
Is CRISPR editing always required?
No. The perturbation should fit the question. Knockdown, transcriptional regulation, overexpression, knockout, or rescue may be considered where technically and biologically appropriate. The page does not promise that every route is available for every organoid type; feasibility is confirmed for the specific model.
Do you perform genome-wide CRISPR screening as part of this service?
Not by default. This service is designed for focused validation of nominated targets. A pooled discovery screen has different library, coverage, selection, and analysis requirements and should be scoped as a separate project rather than assumed from the target-validation package.
Why is DNA verification described as "targeted" rather than whole-exome sequencing?
The immediate question is whether the intended locus contains the expected edit in the material being profiled. Targeted amplicon or Sanger sequencing may answer that question more directly. Broader DNA profiling is considered only when the study rationale requires it; it is not added automatically or used as a substitute for intervention-specific confirmation.
When is bulk RNA-seq enough?
Bulk RNA-seq is appropriate when the primary question concerns an average organoid response and group composition is sufficiently comparable. If the perturbation may alter lineage balance, a rare state, or only a subset of cells, single-cell RNA-seq may better resolve the effect.
Can pathway enrichment prove the target mechanism?
No. Enrichment summarizes whether groups of genes move coherently in relation to annotated pathways. It can support a mechanistic hypothesis, but direct regulation, binding, enzymatic activity, or target engagement requires a suitable orthogonal experiment.
What controls are usually most important?
The answer depends on the perturbation, but a parental or matched background, a non-targeting or mock control, independent biological preparations, and—where feasible—independent reagents, edited populations, or rescue conditions are common considerations. The project plan states which control addresses which alternative explanation.
What happens if the intervention is confirmed but no RNA or phenotype effect is found?
A reproducible negative result can still narrow the hypothesis. We review whether the target was sufficiently perturbed, whether the observation window matched the expected biology, whether the phenotype was sensitive, and whether organoid heterogeneity reduced power. The report does not relabel an inconclusive result as validation.
Can you correlate RNA-seq results with our existing imaging or functional data?
Potentially, if sample identities, replicate relationships, acquisition rules, and time points can be aligned. We first determine whether the datasets describe the same biological units or only parallel batches. The distinction controls how strongly molecular and phenotype results can be associated.
Case Study: Independent Retinal Organoid Study Separates Molecular and Cellular Effects
Source: Jones MK, Orozco LD, Qin H, et al. Frontiers in Cell and Developmental Biology (2023), Figure 2.[5] This is an independent research example, not a CD Genomics customer case.
Background: Investigators examined the consequences of DRAM2 loss using human pluripotent stem-cell-derived retinal organoids.
Methods: Two independently generated biallelic knockout lines and wild-type organoids were differentiated, matured, profiled by single-cell RNA sequencing, and examined with immunolabeling.
Results: Broad pseudobulk analysis showed little differential expression, yet cell-level analysis identified an additional mesenchymal population in knockout organoids. Marker localization supported the presence and position of that population.
Conclusion: The study shows why target validation should preserve independent models, cell-composition context, and orthogonal phenotype evidence. A weak average expression signal did not mean that the perturbation had no model-level consequence.
Independent study figure: molecular and cell-state consequences of DRAM2 loss in retinal organoids.