Directives de soumission d'échantillons
Connect compounds, organoid models, assay controls, and follow-up decisions in one study map.
Define the Decision Before You Define the Throughput
"Screen these compounds" can describe very different studies. One team may need to compare ten candidate molecules across a broad concentration range. Another may need to test hundreds of compounds at one or two concentrations before confirmation. A third may already have two active agents and need a combination matrix. The number of wells is therefore a consequence of the decision, not the starting point.
For a solution-level view from model selection through functional studies and sequencing readouts, see Organoid Research and Sequencing Solutions.
At project intake, we define the unit that will be ranked: a compound, a dose, a combination, a model, or a condition-specific response. We also define what must remain visible when the rank is calculated. A single endpoint can be appropriate for a focused primary screen, but it should not quietly stand in for growth, morphology, cell death, or tissue-specific function when those dimensions matter to the question.
Primary Screening and Confirmation Answer Different Questions
A primary screen identifies conditions that differ from the chosen controls under one assay configuration. A confirmation study asks whether the observation remains reproducible across a new plate, a concentration series, an orthogonal readout, or an additional organoid model. We plan these stages separately so the initial activity threshold does not become an unsupported claim of mechanism, selectivity, or biological generality.
Best for
- Prioritizing small molecules, biologics, tool compounds, or research-grade natural-product fractions in an established organoid assay.
- Comparing single-agent concentration responses across one or more organoid models.
- Mapping pairwise combinations after the single-agent ranges are understood.
- Adding an early normal-tissue organoid or organ-specific counter-screen to a discovery program.
Not for
- Selecting treatment for an individual patient or predicting patient benefit.
- Replacing pharmacokinetic, animal, regulatory toxicology, or clinical evidence.
- Assigning a mechanism solely from a viability curve or image phenotype.
- Screening an undefined mixture when composition, solvent, concentration basis, and handling history cannot be documented.
Match Model, Format, and Readout as One Assay System
An organoid is not automatically screening-ready because it grows in three dimensions. The useful assay window depends on model identity, passage and recovery state, seeding uniformity, extracellular matrix, plate geometry, treatment timing, and the measurement technology. We review these variables together before the compound layout is locked.
| Design element | Question to resolve | Why it changes the result |
|---|---|---|
| Organoid model | Which tissue, genotype, disease state, donor group, or engineered background represents the research question? | Baseline growth, morphology, pathway state, and compound handling can differ across models. |
| Assay format | Does the model perform reproducibly in the proposed well format and matrix configuration? | Miniaturization changes evaporation, edge effects, matrix volume, liquid handling, and the number of analyzable organoids per well. |
| Exposure design | Is the study asking about acute injury, growth suppression, recovery, or a later functional change? | A concentration–time pair defines the observed state. One late endpoint may hide transient or reversible effects. |
| Primary readout | Will viability, morphology, death, reporter signal, or a model-specific function drive hit calling? | Metabolic activity, organoid number, size, architecture, and cell death are related but not interchangeable measurements. |
| Confirmation route | What independent evidence must a condition pass before it is advanced? | Orthogonal confirmation reduces the chance that optical, chemical, or plate artifacts are ranked as biological hits. |
If the organoid model is still being established, the project may first use our organoid characterization services to document morphology, selected markers, and genomic features. Screening begins only after the acceptance logic is appropriate for the planned readout; it is not inferred from a generic photograph.
Select the Screening Module That Fits Your Library
The page covers four purchasing routes under one service. They can be used separately or assembled into a staged program. The exact plate count, concentrations, replicates, and endpoints are defined after feasibility review rather than advertised as universal specifications.
| Module | Typical design | Core output | Key boundary |
|---|---|---|---|
| Single-agent screening | Focused concentration series or primary screen followed by reconfirmation | Normalized responses, fitted curves where supported, effect summaries, replicate review, and prioritized conditions | A fitted half-maximal value is descriptive for the tested model and assay, not a mechanism or human dose. |
| Combination screening | Pairwise dose matrix, fixed-ratio series, or anchor design with single-agent controls | Response surface, reference-model scores, reproducibility flags, and active concentration regions | "Synergy" depends on the selected reference model and tested dose region; one score is not a universal property of the pair. |
| Natural-product screening | Defined compounds, fractions, extracts, or a staged library with documented concentration basis and vehicle | Primary activity map, optical and solubility flags, confirmation results, and a ranked fraction or compound set | Colored, fluorescent, aggregating, or compositionally variable samples require interference controls and careful lot records. |
| Early toxicity module | Normal-tissue or organ-specific organoid counter-screen using matched or relevant exposures | In vitro viability, morphology, injury, or functional readouts alongside the primary screen | This is discovery-stage research evidence, not a regulatory safety assessment or a prediction of clinical toxicity. |
Single Agents: Preserve the Concentration Range, Not Just the Winner
When a shortlist is already available, a concentration series can estimate potency and maximal effect within the tested window. The plate layout retains vehicle and reference controls, and the analysis keeps failed or weak fits visible. A compound that reaches a strong effect only at a concentration with precipitation, broad cytotoxicity, or unstable imaging should not be ranked as if the curve were clean.
Combinations: Establish the Single-Agent Behavior First
Combination matrices are most interpretable when each component spans a biologically useful range and the single-agent edges are present on the same design. Depending on the hypothesis, analysis may compare the observed surface with highest-single-agent, Bliss, Loewe, or zero-interaction-potency expectations. Because these models encode different assumptions, we report the selected method and the concentration region supporting the result instead of converting all matrices into one unqualified synergy label.
Natural Products: Make Composition and Detection Risk Part of the Design
Natural products can enter as purified molecules, standardized fractions, extracts, or client-defined collections. Intake records should capture lot, extraction or fractionation method, solvent, nominal concentration basis, storage, freeze–thaw history, and visible solubility. Pigment, autofluorescence, quenching, precipitation, or nonspecific membrane effects can distort optical and viability readouts, so a credible route includes blank wells, cell-free interference checks where relevant, and an orthogonal confirmation step.
Control the Plate Before Ranking the Compounds
Organoid screens can fail quietly. Edge evaporation, uneven seeding, matrix differences, focal-plane changes, and insufficient object counts may all create patterns that look biological. The pilot therefore tests whether the assay has enough signal separation and reproducibility to support the intended comparison.
- Negative and vehicle controls: establish baseline response and reveal solvent-dependent effects.
- Positive or reference controls: demonstrate that the assay can detect a known change in the chosen endpoint under the current model conditions.
- Plate-position controls: expose drift, edge effects, dispensing order, and imaging variation.
- Biological and technical replication: separate repeat measurements from independent organoid preparations, donors, passages, or experiments.
- Predefined exclusion rules: identify missing organoids, bubbles, segmentation failures, precipitation, contamination, or wells outside the analyzable range without changing the rule after the rank is seen.
Plate-level summaries may include coefficient of variation, signal-to-background, control separation, replicate correlation, and well-count or object-count coverage. We do not advertise one acceptance threshold for every model. The threshold is specified for the project, tied to the readout, and reviewed during feasibility before it becomes a gate.
Keep optical artifacts, precipitation, and generalized injury from bypassing hit confirmation.
Run a Gated Workflow from Feasibility to Confirmed Hits
The workflow is staged so material is not committed to a large screen before the model and assay show that they can answer the question. Each gate has a defined output and a decision about whether to proceed, revise the design, or stop.
Advance from feasibility to screening and confirmation with explicit review points.
- Research-question and material review. Define the model, compound collection, comparison unit, primary endpoint, desired follow-up, and RUO boundary.
- Assay feasibility. Test culture recovery, seeding behavior, exposure handling, detection range, imaging or plate-reader performance, and representative controls.
- Plate-map and analysis lock. Assign compounds, concentrations, controls, randomization, replicates, exclusion rules, and the primary hit metric before screening.
- Screen execution. Record dispensing, incubation, image acquisition, endpoint measurements, deviations, and plate-level QC.
- Primary analysis and triage. Normalize to the appropriate controls, review plate behavior, fit response models where supported, flag interference, and produce the preliminary rank.
- Confirmation. Retest selected conditions using fresh dilutions, concentration expansion, an orthogonal endpoint, an additional organoid preparation, or a relevant normal-tissue model.
- Follow-up package. Deliver confirmed ranks and limitations, then move selected conditions into sequencing or focused mechanism experiments if requested.
Read Dose–Response Curves in the Context of Organoid Growth
A curve compresses many observations into a few parameters. That is useful only when the model assumptions and the underlying points remain visible. We preserve replicate measurements and fit diagnostics alongside summary quantities such as half-maximal concentration, maximal observed effect, or area under the response curve when the data support them.
Different Baseline Growth Can Change Apparent Sensitivity
Two organoid models may begin with similar signal yet divide at different rates during the exposure window. Conventional endpoint metrics can then mix growth rate with compound effect. When baseline and endpoint measurements permit it, growth-rate-aware metrics can help compare proliferating models. When they do not, the report states that limitation rather than presenting a corrected metric without the required data.
Imaging Can Reveal Activity That a Viability Endpoint Misses
Organoid size, number, texture, lumen structure, compactness, and fragmentation can move before or without a large loss of metabolic signal. High-content imaging can therefore add useful phenotypic dimensions, but those features need stable segmentation and adequate object counts. A morphology change may be biologically informative, an injury pattern, or an imaging artifact; it is not automatically a beneficial response.
Time Is a Design Variable
An early measurement may capture acute injury or signaling-associated morphology, while a later measurement may reflect growth arrest, recovery, or selective survival of a subpopulation. If the research question depends on trajectory, the study can include repeated non-destructive imaging or selected time points. The sampling plan is kept compatible with any terminal viability, staining, or sequencing endpoint.
Separate Biological Activity from Assay Interference
Compound-dependent artifacts are especially important when fluorescence, absorbance, luminescence, or image features drive the rank. Autofluorescence can mimic a reporter signal. Quenching can suppress it. Colored fractions can change absorbance. Aggregates and precipitates can alter focus or segmentation. Generalized cell loss can also produce a strong image signature unrelated to the intended phenotype.
We therefore distinguish three labels in the analysis: observed activity under the primary assay, confirmed biological response after appropriate retesting, and unresolved or interference-prone activity when the evidence cannot separate the two. This avoids the false precision of a single hit column.
| Flag | What may be happening | Useful follow-up |
|---|---|---|
| Signal changes without matching images | Detection chemistry, metabolic state, or compound optical properties may dominate the endpoint | Cell-free control, alternate detection chemistry, direct object counts, or orthogonal imaging |
| Strong morphology with low object count | Segmentation may be operating on fragments, debris, or a small survivor population | Review raw fields, object-quality filters, and an independent death or viability readout |
| Curve driven by one concentration | Precipitation, dispensing error, or an unstable fit may create an apparent inflection | Fresh dilution, expanded concentration range, and replicate confirmation |
| Plate-position pattern | Evaporation, temperature, dispensing sequence, or imaging drift may confound treatment | Randomization, perimeter controls, plate-map review, and repeat execution |
Treat Early Toxicity as a Research Counter-Screen
Early toxicity work can add a second research question: does a compound that changes the primary model also disrupt a relevant normal-tissue or organ-specific organoid under the tested in vitro conditions? The answer can help prioritize which compounds deserve more detailed follow-up, but it does not establish a clinical safety margin.
The module may combine viability and morphology with an agreed functional or injury-associated readout appropriate to the model. Examples include barrier-related behavior in epithelial systems, contractility or rhythm-associated measurements in cardiac models, or selected injury markers in liver or kidney organoids. Protein measurements, when included, remain focused assay readouts within the organoid screen; this page does not expand into a proteomics service.
Interpretation keeps exposure comparability, model maturity, solvent tolerance, assay duration, and biological replication visible. A lack of change in one organoid model means only that no change was detected in that model, endpoint, and tested exposure range. It cannot be generalized to whole-organism safety, metabolism, or regulatory toxicology.
Move Confirmed Hits into Sequencing and Mechanism Studies
Phenotypic screening answers which conditions merit attention. It does not, by itself, explain why they changed. Confirmed hits can be advanced into a focused molecular study after the screen has reduced the condition space.
For a compact set of matched treatments, our organoid drug response transcriptomics services can compare treatment-associated expression programs. Higher-condition studies may use the Drug-seq service, while deeper coverage for a smaller set may use the mRNA sequencing service. When a bulk signal may reflect different cell states, single-cell RNA sequencing can be considered if the material and question justify it.
The handoff preserves compound identity, concentration, exposure time, plate and batch, organoid model, response metric, and confirmation status. This prevents the sequencing comparison from becoming disconnected from the screen that selected the samples. Sequencing remains an optional follow-up; it is not required for every hit and is not used to rescue an uninterpretable primary assay.
Receive an Auditable Screening Package
The deliverable is designed for scientific review and the next research decision. It retains the evidence behind each rank rather than presenting only a polished shortlist.
- Final experimental design, plate maps, compound and solvent metadata, control definitions, and deviation log.
- Raw or agreed processed endpoint data, image references, well-level QC, and exclusion records.
- Normalized response matrices, replicate summaries, fitted curves and diagnostics where applicable.
- Combination response surfaces and clearly named reference-model outputs when included.
- Interference, precipitation, low-object-count, edge-effect, and confirmation flags.
- Primary and confirmed hit tables with the ranking rule and limitations stated.
- Optional matched-sample manifest for downstream organoid sequencing.
What We Need to Scope the Study
Provide the organoid source and current culture status, the number and type of compounds or fractions, available concentration and solvent information, preferred exposure windows, the scientific decision the screen must support, and any required readouts. If you already have pilot data, share the raw plate or image summaries rather than only the final curve. We use that information to determine whether the existing assay can be transferred, needs a feasibility phase, or should be redesigned before material is consumed.
Illustrative Organoid Drug Screening Results
These mock outputs demonstrate three distinct ways to interrogate a screen: concentration-dependent activity, pairwise interaction, and an early toxicity counter-screen. They contain no customer observations and establish no universal cutoff or expected result. The final visualization package is selected only after the study's model, plate geometry, assay controls, and usable measurements are known.
Illustrative example: interpret the curve together with replicate and image behavior.
Single-Agent Response Atlas
A response atlas can place normalized endpoint values beside concentration–response fits and representative organoid images. Compounds are ranked only after fit quality, maximal observed effect, replicate consistency, and visual flags are reviewed. Failed fits and incomplete concentration coverage stay in the table rather than being silently removed.
Illustrative example: localize combination activity within the tested dose matrix.
Combination Response Surface
A pairwise matrix preserves both single-agent edges and the combined response at each concentration pair. The report can show observed response, model-specific excess response, replicate spread, and the dose region driving the score. A localized result is not described as a property of every dose or every organoid model.
Illustrative example: compare primary activity with a relevant normal-tissue counter-screen.
Early Toxicity and Selectivity View
A paired view can compare the primary model with a normal-tissue or organ-specific organoid using matched exposure records. Viability, morphology, and selected functional readouts remain separate columns. The comparison supports discovery prioritization only and does not claim a clinical or regulatory safety margin.
Organoid Drug Screening FAQs
Can you start with organoids that we already maintain?
Yes, subject to feasibility review. We need the model identity, derivation and passage history, recent culture performance, matrix and medium, contamination status, and any existing characterization or screening data. A transfer or pilot phase may be needed before a larger plate campaign.
Do you require a high-throughput format?
No. A focused set of candidates with deeper concentration coverage and more readouts may be more informative than a large primary screen. Throughput follows the research decision, available organoid material, and assay performance.
Can single agents and combinations be included in the same project?
Yes, but combinations are usually planned after useful single-agent ranges are established. The combination stage needs single-agent controls and a predefined interaction model so the joint response can be interpreted within the tested concentration region.
Can you screen natural-product extracts or fractions?
Potentially. The project needs documented lot, extraction or fractionation method, solvent, concentration basis, storage history, and visible solubility. Colored, fluorescent, quenching, precipitating, or complex samples may require cell-free controls and an orthogonal confirmation assay.
Which readout is best for an organoid screen?
There is no universal best endpoint. Metabolic viability is efficient, imaging adds organoid-level and morphology information, death markers distinguish injury, and model-specific functions can address a focused hypothesis. We select the smallest readout set that can support the intended decision.
Does an IC50 value prove that a compound is potent in vivo?
No. It describes a fitted response within the tested in vitro model, endpoint, exposure window, and concentration range. It does not account for absorption, distribution, metabolism, whole-organism exposure, or clinical dosing.
How do you handle organoid models with different growth rates?
We preserve baseline and endpoint context and may use growth-rate-aware metrics when the design provides the required measurements. Otherwise, model-to-model differences are interpreted with explicit growth and assay limitations.
What does "early toxicity" mean on this page?
It means a research counter-screen in a relevant normal-tissue or organ-specific organoid using agreed in vitro exposures and endpoints. It is not a regulatory toxicology package, a whole-organism safety assessment, or a prediction of human adverse events.
Can sequencing be added after the screen?
Yes. Confirmed conditions can enter focused bulk RNA sequencing, higher-throughput expression profiling, single-cell analysis, or another justified sequencing route. The sample plan must preserve treatment, timing, plate, batch, and model metadata.
What happens if the pilot does not produce a usable assay window?
We report the failure mode and recommend whether to adjust seeding, timing, format, controls, or readout. A large screen should not proceed simply because compounds and plates are available.
Can these results be used to select a therapy for a patient?
No. This service is for research use only. It does not provide diagnosis, patient stratification, therapeutic selection, clinical decision support, or clinical-trial testing.
Case Study: Independent Screen Shows Why Activity and Viability Are Not Equivalent
Source: Betge J, Rindtorff N, Sauer J, et al. Nature Communications (2022), Figure 3.[1] This is an independent research example, not a CD Genomics customer case.
The study used high-throughput image-based profiling to examine colorectal cancer organoids exposed to more than 500 small molecules. Across the broader experiment, the investigators quantified approximately 5.5 million individual organoids and extracted morphology features alongside viability-associated measurements.
What Figure 3 Contributes
Figure 3 compares morphology-based drug activity with viability classification and shows representative organoid phenotypes. Some perturbations produced a reproducible morphological signature without meeting the study's lethal-response definition. The figure therefore demonstrates a practical screening lesson: a compound can be active in a phenotypic space even when a single viability endpoint does not call it cytotoxic.
How This Informs Service Design
For a discovery screen, morphology and viability should be interpreted as complementary evidence rather than interchangeable labels. A morphology-active condition may deserve confirmation because it changes architecture, size, texture, or cellular organization. It may also represent stress, nonspecific injury, or an analysis artifact. The next step is to review raw images, object counts, replicate behavior, concentration dependence, and an orthogonal endpoint before advancing the condition.
The published figure is reproduced for planning reference under the article's Creative Commons Attribution 4.0 International License. The service does not copy the study's thresholds into a customer project; assay gates are defined around the selected model and readout.
Independent study figure: morphology-based activity can add information beyond a terminal viability call.