<img alt="" src="https://secure.soil5hear.com/223550.png" style="display:none;">
Contact Us Select Models
Select Models
1855764562

Matched multi-omic dataset

Every layer measured on the same donor's tumor, from DNA to drug response

Whole-exome and RNA sequencing, whole-cell and cell-surface proteomics, clinical history and functional response, matched across 1,500+ low-passage patient-derived tumor models.

1,500+ sequenced models1,000+ with proteomics600+ with surface data

Overview

A dataset is only as useful as the link between its layers, and ours share a donor

Most multi-omic resources assemble their layers from different samples, cohorts or cell lines. That is workable for generating a hypothesis, but it leaves a bridging assumption under every attribution: that the tumor you sequenced behaves like the tumor that was dosed. When those are different tumors, the assumption cannot be tested.

We measure every layer on the same low-passage patient-derived models. Sequencing, whole-cell and cell-surface proteomics, IHC, clinical annotation and in vivo and ex vivo response are matched by donor, so a molecular feature and the functional result it may explain come from one tumor rather than two.

The dataset supports three uses: enriching public and internal reference data, training and validating your own computational models, and selecting models for a study before anything is dosed. It keeps growing, because any model in the bank can be profiled on request.
96309378

At a glance

RNA-Seq
1,615 models
Whole-exome
1,599 models
Whole-cell protein
1,006 models
Surface proteomics
607 models
View study package

Data layers

Six layers, each matched to the same model

Each layer below is held per model and linked by donor. Not every model carries every layer yet; the coverage table shows where each stands, and missing layers can be generated.

  • Whole-exome sequencing (WES)

    Mutations in protein-coding regions plus copy-number changes, called by five parallel variant callers after mouse reads are removed.

  • RNA sequencing (RNA-seq)

    Gene and isoform expression and fusion detection from polyadenylated transcripts, quantified after stromal reads are stripped out.

  • Whole-cell proteomics (WCP)

    Peptide and protein abundance across the proteome by ultra-deep DIA mass spectrometry, covering tumor and surrounding stroma.

  • Cell-surface proteomics (CSP)

    Membrane-localized proteins measured from a separately enriched membrane fraction, so surface abundance is measured rather than inferred.

  • Clinical and treatment history

    Tumor classification, stage, biomarkers, harvest site and prior therapies, recorded in the order the patient received them.

  • Functional drug response data

    Tumor growth inhibition from in vivo studies and response from ex vivo assays, including standard-of-care agents.

Coverage

Where each layer stands, by tumor type

Model counts for the most represented tumor types in the bank. Low counts are not a ceiling, because any layer can be generated for any model you select.
Tumor typeRNA-SeqWESWhole-cellCell surface
Colorectal21720920070
NSCLC208210167136
Sarcoma148145On request22
Breast13813812965
Head and neck13013011871
Ovarian90896234
Pancreatic86867966
AML777512On request
Melanoma50504739
All tumor types1,6151,5991,006607

Model counts per layer from the September 2026 data platform overview. Figures grow as profiling continues; any layer can be generated on request.

2174551157

Get Started

Access the Champions Data Room

For partners ready to explore the depth of our data, Champions provides secure access to our Data Room. Here you can evaluate sample datasets, review enrichment metrics, and see how our cell surface proteomics integrate with broader multiomic profiles.

  • Discover how surface expression correlates with treatment response.

  • Explore new target opportunities validated in real human tumors.

  • Assess the quality and interpretability of our data for your own programs.

Clinical annotation

Select models the way a trial selects patients, not by model name

Each model carries the record of the patient it came from. That lets you filter by line of therapy, stage or biomarker, and see how the source tumor behaved before it became a model.

Disease context

Primary cancer type, subtype, histology, grade and stage, with whether the sample came at first diagnosis, recurrence or relapse.

80% late stage

Prior therapy

Treatments the patient received before the sample was taken, in order, so resistance can be studied in tumors that have already been treated.

65% pretreated

Specimen origin

Anatomical harvest site and whether the tumor was primary or metastatic, so you can match the biology of the site that matters to you.

55% metastatic

Biomarkers

Microsatellite instability, tumor mutational burden, homologous recombination deficiency and HPV status, plus triple-negative breast cancer subtyping where it applies to the model.

Filterable per model

536688016

On Demand

An unprofiled model is an option,not a gap in the data

The counts above are what has been generated so far, not what is available. Molecular layers are produced on request across the bank, so model selection can follow clinical and pharmacological fit to your question rather than which models happen to carry data already.

Each layer you add is generated with the same protocols and analysis pipelines as the rest of the dataset. New data is directly comparable with what is already there, and it stays matched to the response data held on the same model.

FAQ

What data teams ask before licensing

Questions that come up when a computational team evaluates the dataset for its own models.

Every passage is a chance for a tumor to drift from the patient it came from. Low-passage models keep more of the source tumor's heterogeneity and its microenvironment-shaped biology, which matters when the question is how a patient population will respond.

Into your own environment, as curated cohort files with per-model clinical annotation and documented processing methods. Cohorts are scoped by indication or data layer, and refreshed data can be included as profiling continues during the license term.

Sequencing is aligned to GRCh37 after mouse reads are removed. Expression is quantified with RSEM and fusions with FusionCatcher, and variants are merged from five callers including MuTect and Strelka2. Full methods are shared with licensed data.

Yes. Tell us the indication, prior therapy or biomarker you are selecting on, and we will return the matching models with the layers each already carries and what would still need to be generated for your study.

Related pages

Where the dataset is put to work

These pages show how the same matched layers are used to answer specific program questions.
1855764562

Check coverage

Send the indication and the layers you need

Tell us the indication or molecular feature you are selecting on, the layers you need and whether you want data for your own models or an analysis run for you. We will reply with the matching models and their current coverage.

  • A data scientist replies to every request personally

  • Matching models and layer coverage included

  • CDA handled before any program detail is shared