Sample Submission Guidelines
Connect Cell Identity, Regulation, and Function in One Study
Tissue averages can hide rare populations. A single molecular layer can identify a pattern without explaining what controls it.
Single-cell RNA sequencing can reveal cell types and expression states, but similar RNA profiles may arise through different regulatory routes. Chromatin accessibility can point to active regulatory regions. Cell-surface proteins can refine immune phenotypes, while T-cell and B-cell receptor sequences can connect a state to clonal expansion.
We design the study around the decision you need to make. That may mean measuring RNA and chromatin in the same nucleus, pairing RNA with surface proteins, adding immune receptor sequencing, or integrating compatible measurements across samples.
AI-assisted methods help compare thousands of features and map cells to appropriate references. Our scientists review quality, annotation evidence, batch structure, biological consistency, and uncertainty before a result is reported.
Questions we help you answer
- Which cell types and states are present?
- Which states change across groups or time points?
- Which regulatory programs support those states?
- Do surface proteins refine the RNA-defined identity?
- Which clonotypes expand within a specific state?
- Which findings remain consistent across donors?
Single-Cell Multi-Omics for Specific Research Decisions
Each scenario connects a biological question to suitable experiments, integration methods, quality tests, and usable outputs.
Resolve Tumor and Immune Microenvironment Heterogeneity
Your question: Which malignant, immune, and stromal states differ across tumors, models, treatment groups, or tissue regions?
Data and experiments: scRNA-seq provides a broad cell-state map. CITE-seq can refine surface phenotype, scATAC-seq can reveal regulatory differences, and immune repertoire sequencing can connect lymphocyte states to clonotypes. We can generate one or several layers according to sample suitability.
How AI helps: Reference mapping and multimodal neighbor models help organize related cells while preserving sample identity. Candidate malignant programs, immune states, and cell interactions are then reviewed against known markers and the study context.
How we test the result: We check donor support, batch balance, annotation evidence, cell abundance, and within-state changes. A pattern found in many cells from one donor is not treated as cohort-wide evidence.
What we deliver: You receive cell and state maps, group comparisons, marker tables, regulatory or interaction results, donor-level summaries, and candidates for focused follow-up.
Published evidence: Sivakumar et al. combined matched tumor and blood scRNA-seq, CITE-seq, and immune receptor data from pancreatic cancer samples with public datasets. Their study shows how several single-cell layers can distinguish immune microenvironment patterns that would be difficult to resolve from averaged tissue data.
Separate Drug-Response States From Resistant Cell Populations
Your question: Does a treatment change the number of cells in a state, change the state itself, or select a pre-existing resistant population?
Data and experiments: We can compare treated and untreated samples, sensitive and resistant models, dose groups, time points, or perturbations. RNA, chromatin, surface protein, and genotype-aware measurements can be combined when the question requires them.
How AI helps: AI-assisted models can match comparable states across conditions and prioritize features that change together across modalities. They can also help separate a shift in cell composition from a molecular change within the same state.
How we test the result: Comparisons are made at the sample level when possible. We review donor, model, treatment, time, and batch structure and test whether the finding depends on one annotation or integration choice.
What we deliver: You receive response-associated states, differential features, candidate regulatory programs, state-transition evidence, and a practical plan for orthogonal or perturbation-based validation.
Published evidence: Wegmann et al. integrated single-cell ex vivo drug profiling with DNA, RNA, and protein measurements from relapsed or refractory acute myeloid leukemia samples. The work illustrates how cell-resolved molecular evidence can distinguish innate and acquired resistance patterns.
Study Development, Differentiation, and Disease Progression
Your question: Which intermediate states connect an early population to later cell identities, and which regulatory programs change along the path?
Data and experiments: RNA measurements define changing expression states. Paired RNA and chromatin measurements can connect those states to accessible regions and candidate regulators. Time points, anatomical regions, or experimentally defined stages strengthen interpretation.
How AI helps: Trajectory and latent-state models order cells by molecular similarity and compare candidate branches. Regulatory models connect transcription factors, accessible regions, and target-gene expression when the data support those links.
How we test the result: We compare the inferred order with collection time, known markers, replicate samples, and the second molecular layer. A trajectory is a model-based research interpretation; it is not direct proof of lineage history.
What we deliver: You receive state and branch maps, time- or stage-associated features, candidate regulators, supporting and conflicting evidence, and follow-up experiments.
Published evidence: Zuo et al. generated paired RNA and chromatin profiles from the same nuclei across human retinal development. The study linked developmental states to trajectories, transcription factors, and gene regulatory networks.
Connect Immune Clonotypes With Functional Cell States
Your question: Which T-cell or B-cell clones expand, and what expression or surface-protein states do those clones occupy?
Data and experiments: scTCR/BCR-seq can be paired with gene expression and, when suitable, surface-protein measurements. Samples may represent tissues, blood-derived cells, treatment groups, time points, or research models.
How AI helps: AI-assisted annotation organizes immune states and helps compare clonotype distributions across those states. Shared receptor features, clonal expansion, state enrichment, and cross-sample recurrence can then be examined.
How we test the result: We report the number of supporting cells and samples, review receptor quality and chain pairing, and avoid treating one expanded clone as a general group effect.
What we deliver: You receive clonotype tables, state-linked expansion maps, diversity summaries, cross-sample comparisons, and candidate clones or states for research follow-up.
Published evidence: ECCITE-seq demonstrated joint measurement of transcriptomes, surface proteins, clonotypes, and perturbation readouts. It provides a technical foundation for connecting immune receptor identity with cellular phenotype in a multimodal design.
Choose Molecular Layers That Change the Next Decision
We recommend technologies according to the sample, biological question, required resolution, and evidence gap.
| Service Technology | Evidence Added | Typical Role in the Study |
|---|---|---|
| Single-Cell RNA Sequencing | Gene expression by cell | Cell types, cell states, rare populations, pathways, and group comparisons |
| Single-Cell ATAC-Seq | Accessible chromatin by cell or nucleus | Regulatory states, motif activity, candidate transcription factors, and linked regions |
| Single-Cell ATAC and RNA Profiling | RNA and chromatin from the same nucleus | Direct pairing of expression state and regulatory evidence without cross-cell matching |
| CITE-seq | Gene expression and selected surface proteins | Immune phenotyping, protein-supported annotation, and state refinement |
| Single-Cell TCR/BCR Sequencing | Immune receptor clonotypes linked to cell states | Clonal expansion, repertoire diversity, state distribution, and cross-sample tracking |
| Spatial Multi-Omics Sequencing Services | Tissue location and neighborhood context | Optional follow-up to locate important states, interactions, or resistant niches in tissue |
| Multi-Omics Services | Additional molecular evidence across the study | Focused genomic, epigenomic, transcriptomic, or other omics support |
| Bioinformatics Services | Processing and interpretation of existing data | Quality review, harmonization, public-data integration, and reproducible reporting |
Not every project needs every modality. Adding a data layer is useful only when it can resolve a specific uncertainty, such as whether an RNA-defined state has a distinct regulatory program, surface phenotype, clonotype pattern, or tissue location.
Start With Samples, Existing Data, or a Hybrid Design
| Project Entry | What You Provide | How We Support the Study |
|---|---|---|
| Sample-to-Insight | Suitable cells, nuclei, tissue, blood-derived cells, organoids, or research models; study groups and metadata | We review feasibility, design and perform the agreed single-cell experiments, process each modality, integrate the evidence, and deliver scientist-reviewed findings. |
| Data-to-Insight | FASTQ files, feature-barcode or count matrices, cell-level metadata, Seurat objects, AnnData objects, or compatible public datasets | We review data quality and study structure, process or harmonize the data, perform multimodal analysis, and report reproducible results and limits. |
| Hybrid Study | Existing data plus biospecimens or a planned experiment for the missing evidence | We identify the gap, generate a focused new molecular layer, and connect it to the existing single-cell study. |
Keep Batch Correction From Erasing Real Biology
Integration is successful only when it reduces unwanted technical variation and preserves meaningful differences among samples and states.
- Cell and nucleus quality: We review feature detection, library complexity, ambient signal, multiplets, and low-quality profiles using data-appropriate measures.
- Sample structure: Donor, group, batch, site, treatment, time, and processing variables remain visible throughout the analysis.
- Annotation evidence: Labels are supported by markers, reference sources, multimodal agreement, and biological context.
- Ambiguity: Mixed, transitional, or weakly supported states are flagged instead of being forced into a confident label.
- Biological preservation: We check whether expected donor, group, lineage, and state differences remain after integration.
- Cross-modal agreement: RNA, chromatin, protein, or receptor evidence is compared where a direct or biologically justified link exists.
- Sample-level support: Conclusions are not based only on a large number of cells from one sample.
- Sensitivity review: Important findings are checked under reasonable filtering, annotation, integration, and model choices.
Benchmarking work has shown that no integration method is best for every dataset. We choose and test methods according to modality, batch structure, biological question, and the kind of variation that must remain visible.
From Study Question to Tested Multimodal Evidence
One connected workflow keeps wet-lab design, modality-specific quality control, integration, interpretation, and validation focused on the same decision.
Step 1 - Define the research decision: We clarify the biological contrast, unit of replication, required cell resolution, and how the result will guide the next experiment.
Step 2 - Review samples and existing data: We examine sample suitability, expected cell types, preservation, group balance, metadata, batch structure, and available data objects.
Step 3 - Generate the selected modalities: When biospecimens are included, we perform the agreed single-cell or single-nucleus experiments and keep sample identity traceable.
Step 4 - Apply modality-specific quality control: RNA, chromatin, protein, and receptor data are assessed with suitable measures before integration.
Step 5 - Integrate compatible evidence: We select a method that can align related cells or features while preserving relevant donor, group, lineage, and state differences.
Step 6 - Interpret cell states and mechanisms: We examine annotations, differential abundance, within-state changes, regulatory programs, trajectories, interactions, or clonotypes as supported by the design.
Step 7 - Test cross-sample support: Important results are reviewed by sample, group, batch, and reasonable method choices. Confidence and conflicting evidence are reported.
Step 8 - Plan focused follow-up: We prioritize candidates for targeted assays, perturbation studies, additional cohorts, or spatial localization.
What We Review Before the Study Starts
Sample and metadata review comes before a fixed protocol recommendation because tissue, preservation, cell recovery, and study structure affect the design.
- Research question, comparison groups, and the decision the study should support
- Sample type, tissue source, preservation, expected cell or nucleus recovery, and known quality concerns
- Donor, model, treatment, dose, time point, tissue region, collection site, and batch information
- Available scRNA-seq, scATAC-seq, CITE-seq, immune receptor, multiome, or compatible public data
- FASTQ files, count matrices, feature-barcode matrices, Seurat objects, AnnData objects, and cell-level metadata as applicable
- Expected rare populations, important markers, prior findings, and suitable reference datasets
- Available material or cohorts for technical, biological, or spatial follow-up
After feasibility review, we provide a study-specific recommendation for modalities, sample handling, replication, metadata, quality checks, and follow-up options.
Deliverables You Can Review, Reuse, and Test
- Experimental design and sample-feasibility review
- Modality-specific quality report
- Retained cell or nucleus summary by sample
- Processed and harmonized cell-level data objects
- Cell-type and cell-state annotations
- Annotation evidence, confidence, and ambiguity flags
- Marker and differential-feature tables
- Differential abundance and within-state comparisons
- Multimodal factors, linked features, or regulatory programs
- Trajectory, interaction, or clonotype results as applicable
- Donor, batch, and sensitivity assessment
- Publication-ready figures and result tables
- Reproducible methods, scripts, and data-object guidance
- Limitations and focused follow-up recommendations
Close the Gap Between a Cell Map and a Testable Mechanism
When a project is limited to existing data, a key uncertainty may remain unresolved. RNA may define an important state without showing its regulatory basis. A receptor dataset may show clonal expansion without explaining the phenotype of those cells.
CD Genomics can connect study design, single-cell and single-nucleus experiments, multimodal bioinformatics, AI-assisted pattern review, and scientist interpretation. This makes it possible to add the evidence that the biological question actually needs.
We deliver both the integrated result and the evidence behind it: sample-level support, annotation sources, quality checks, sensitivity results, limitations, and a focused next-study plan.
Research boundary
Cell states, trajectories, interactions, and regulatory links are research interpretations shaped by sample quality, study design, data modality, and model choice. We do not guarantee that every modality will agree or that every candidate mechanism will be confirmed in follow-up experiments.
References
- Hao Y, Hao S, Andersen-Nissen E, et al. Integrated analysis of multimodal single-cell data. Cell. 2021.
- Luecken MD, Büttner M, Chaichoompu K, et al. Benchmarking atlas-level data integration in single-cell genomics. Nature Methods. 2022.
- Stoeckius M, Hafemeister C, Stephenson W, et al. Simultaneous epitope and transcriptome measurement in single cells. Nature Methods. 2017.
- Mimitou EP, Cheng A, Montalbano A, et al. Multiplexed detection of proteins, transcriptomes, clonotypes and CRISPR perturbations in single cells. Nature Methods. 2019.
- Wegmann R, Bonilla X, Casanova R, et al. Single-cell landscape of innate and acquired drug resistance in acute myeloid leukemia. Nature Communications. 2024.
- Sivakumar S, Jainarayanan A, Arbe-Barnes E, et al. Distinct immune cell infiltration patterns in pancreatic ductal adenocarcinoma exhibit divergent immune cell selection and immunosuppressive mechanisms. Nature Communications. 2025.
- Zuo Z, Cheng X, Ferdous S, et al. Single cell dual-omic atlas of the human developing retina. Nature Communications. 2024.
Example Single-Cell Multi-Omics Report
The report connects the biological finding to sample-level support, multimodal evidence, and quality checks. It is designed to help you decide what should move into the next experiment.
A project-specific report may include cell maps split by donor and group, cell-type proportions, annotation evidence, RNA and chromatin feature links, surface-protein support, regulatory programs, trajectory or interaction results, clonotype overlays, sensitivity checks, and prioritized follow-up candidates.
AI-Assisted Single-Cell Multi-Omics FAQs
1. Can the project begin with samples rather than existing data?
Yes. We can review sample suitability, recommend the molecular layers, perform the agreed experiments, process each modality, and integrate the results. The recommendation depends on tissue, preservation, expected cell recovery, study groups, and the biological question.
2. Can you work with data generated elsewhere?
Yes. We can start from raw files, count matrices, feature-barcode matrices, Seurat objects, AnnData objects, or compatible public datasets. We first review data quality, sample identity, metadata, and batch structure.
3. Does every study need several single-cell assays?
No. More modalities do not automatically produce a better answer. We add a layer only when it can resolve an important uncertainty, such as regulation, surface phenotype, clonality, or tissue location.
4. What does AI do in the project?
AI-assisted methods can support reference mapping, annotation, multimodal integration, pattern comparison, and candidate prioritization. Our scientists choose the analysis strategy, inspect supporting evidence, review uncertainty, and interpret the result in the study context.
5. How do you handle batch effects?
We keep donor, group, batch, and processing variables visible, select an appropriate integration method, and check whether expected biology remains after correction. We also compare important findings across reasonable processing choices.
6. How do you avoid overconfident cell labels?
Annotations are reviewed against marker expression, suitable references, sample context, and other molecular layers when available. Mixed, transitional, or weakly supported populations can receive an ambiguity flag instead of a forced label.
7. Can RNA and chromatin be measured in the same cell?
Compatible single-nucleus multiome designs can measure RNA and chromatin accessibility from the same nucleus. This removes the need to match those two layers across different cells, although donor and batch effects still require review.
8. Can spatial omics be added?
Yes, as a separate follow-up when tissue location matters. Spatial profiling can test where an important cell state, interaction, or resistant niche occurs. It is not required for every single-cell project.
9. Does a trajectory prove lineage history?
No. A trajectory orders cells according to measured molecular similarity and model assumptions. Time points, lineage tracing, perturbation, imaging, or other experiments may be needed to test the proposed path.
Published Case Study
Independent Research Highlight
Connecting Gene Expression and Chromatin Accessibility During Human Retinal Development
This publication is an independent research example. It is not a CD Genomics customer project.
Background
Developmental cell states change in both gene expression and chromatin regulation. The researchers asked how these layers could be connected across time and anatomical location in the developing human retina.
Methods
Zuo et al. profiled 315,177 nuclei from 28 retinal tissues collected from 14 human embryos and fetuses between post-conception weeks 8 and 23. After quality control, 226,506 nuclei remained. RNA and accessible chromatin were measured from the same nuclei. Figure 1 presents the study design, cell classes, and agreement between expression and chromatin-derived gene scores.
Results
The researchers identified 61 cell types, including five precursor groups. Figure 2 shows estimated progenitor-state flow, latent time, expression modules, and a candidate regulatory network. Figure 4 examines the path from amacrine-cell progenitors to committed subclasses. Together, the results connected changing cell states with candidate transcription factors and gene regulatory programs.
Why It Matters
The study demonstrates the value of measuring two molecular layers in the same nucleus. It also shows how time, anatomical region, cell annotation, trajectory models, and regulatory evidence can be reviewed together rather than treated as separate outputs.
Conclusion
A strong single-cell multi-omics study connects experimental design, modality-specific quality control, cell-state interpretation, and cross-layer evidence. The inferred trajectories and regulatory links remain research models that require biological review and, when important, experimental follow-up. Results from this publication do not predict the outcome of another project.
Open-access note: The article is licensed under the Creative Commons Attribution 4.0 International License. This page describes the study without reproducing its figures.
Reference
- Zuo Z, Cheng X, Ferdous S, et al. Single cell dual-omic atlas of the human developing retina. Nature Communications. 2024.
Selected Publications
These independent publications provide research foundations for multimodal integration, quality-aware harmonization, surface-protein measurement, immune clonotype analysis, drug-response research, and paired RNA-chromatin studies. They are not presented as CD Genomics customer projects.
- Hao Y, Hao S, Andersen-Nissen E, et al. Integrated analysis of multimodal single-cell data. Cell. 2021.
- Luecken MD, Büttner M, Chaichoompu K, et al. Benchmarking atlas-level data integration in single-cell genomics. Nature Methods. 2022.
- Stoeckius M, Hafemeister C, Stephenson W, et al. Simultaneous epitope and transcriptome measurement in single cells. Nature Methods. 2017.
- Mimitou EP, Cheng A, Montalbano A, et al. Multiplexed detection of proteins, transcriptomes, clonotypes and CRISPR perturbations in single cells. Nature Methods. 2019.
- Wegmann R, Bonilla X, Casanova R, et al. Single-cell landscape of innate and acquired drug resistance in acute myeloid leukemia. Nature Communications. 2024.
- Sivakumar S, Jainarayanan A, Arbe-Barnes E, et al. Distinct immune cell infiltration patterns in pancreatic ductal adenocarcinoma exhibit divergent immune cell selection and immunosuppressive mechanisms. Nature Communications. 2025.
- Zuo Z, Cheng X, Ferdous S, et al. Single cell dual-omic atlas of the human developing retina. Nature Communications. 2024.
