At a glance
- Panel designs
- 4
- Core tumor types
- PDAC, CRC, NSCLC
- Models available
- XX
- Variable tested
- One per panel
KRAS panels for in vivo studies
Pick the KRAS models that answer your question before you dose
Four KRAS panel designs, each built around one question, from treatment history in G12D pancreatic cancer to activity across tumor types, or build your own, with daraxonrasib available as the benchmark.
Four panel designsDaraxonrasib benchmark optionClinically annotated models
Overview
Allele status alone does not tell you which models will answer your question
A KRAS-mutant label groups tumors that behave very differently. Allele, tumor type, co-mutations, pathway activity and prior treatment all shape response, and a model set chosen on mutation status alone can leave a study unable to separate a real signal from a model that was never going to respond. Daraxonrasib has raised the benchmark that new KRAS programs are compared against.
Our KRAS panels are built around that question. You choose a panel design and size, or build your own, and we select models with you using allele, tumor type, co-mutation context and measured protein biology. Daraxonrasib can be included as a benchmark arm, and your compound can run alone, in combination or head-to-head with it.
Each panel holds one factor constant and varies another, such as treatment history, allele or tumor type. Study groups and sampling timepoints are set with you, and the final model list is confirmed against availability and external shareability before you commit.
Panel designs
Four ways to build a panel around your question
Start from the design closest to your program. Each is an example panel that sets the scope; final model selection stays yours, and any design can be adjusted for allele, indication, comparator, or combination arm.
PDAC panel with a G12D focus
A focused PDAC panel for G12D-directed programs. Models are chosen to reflect different patient treatment histories, from treatment-naive to pretreated, so you can see where activity holds. Comparator alleles such as G12R or G12V can be added to test allele specificity.
Use it to compare your asset with daraxonrasib, or to test a combination, in the tumor type where the clinical conversation is most active.
Best suited to
G12D-targeted assets and combinations
Questions about treatment history
Programs testing allele specificity
Panel sizes
Size the panel to the decision it has to support
Three starting sizes let us scope quickly once you choose a direction. The right size depends on whether you need fast proof of concept, a stronger translational readout, or coverage for portfolio strategy.
Early entry
Lean panel
The Lean panel answers one focused question quickly, such as whether your compound shows activity in G12D PDAC models compared with daraxonrasib. It keeps the first study small while still giving a benchmarked result.
Use it to test a hypothesis before committing to a larger design, then expand the model set once the first readout shows where the signal sits.
Best suited to
- Fast proof of concept
- Early customer entry
- Single-question study designs
- Models
- 8
- Design
- One question
Recommended
Recommended panel
Twelve models add breadth across alleles, co-mutation contexts, or tumor types while keeping the study focused. It is our recommended starting point for a stronger translational readout.
Efficacy groups and pharmacodynamic or biomarker groups can be added where the question needs them, so response and mechanism are read side by side in the same study.
Best suited to
- Default for translational readouts
- Response and mechanism together
- Stronger comparison across alleles
- Models
- 12
- Design
- Translational
Portfolio scale
Expanded panel
Twenty to twenty-four models support cross-indication comparisons and subgroup analysis that a smaller panel may not resolve. The set is built from PDAC, CRC, and NSCLC models where the science justifies each inclusion.
It suits late-stage assets and portfolio strategy, where the decision depends on knowing which tumor types and alleles to prioritize next. Deeper omics and longer sampling can be added when they change the next decision.
Best suited to
- Late-stage assets and portfolio strategy
- Cross-indication and subgroup comparisons
- Programs choosing where to invest
- Models
- 20-24
- Design
- Portfolio
Not sure which panel fits your program?
Tell us your allele, indication, and the decision the study has to support. A KRAS scientist will help you choose the design.
Study path
From your question to a benchmarked readout in five steps
Each step ends in something that either happened or did not, so you know where the study stands and what decision comes next.
- 01
Define the question
A KRAS scientist and your team agree what the study must decide, such as efficacy, allele specificity, combination, durability or a biomarker hypothesis.Exit criterion — a written question and decision criterion are agreed.
- 02
Select the models
We propose a model set by allele, tumor type, co-mutation context and measured protein biology, then confirm availability and external shareability.Exit criterion — the final model list is confirmed in writing.
- 03
Design the arms
Daraxonrasib can be included as a benchmark arm. Your compound runs alone, in combination or head-to-head, with sampling timed to the question.Exit criterion — arms, endpoints and sampling schedule are signed off.
- 04
Run and sample
Efficacy and pharmacodynamic groups are dosed, with samples taken at the timepoints the design calls for.Exit criterion — every group reaches its planned endpoint and sampling point.
- 05
Interpret and decide
Our scientists read response alongside molecular data and propose mechanism, biomarker and combination hypotheses, plus the follow-on study design they point to.Exit criterion — you receive a readout tied to the original question.
Why Champions
Clinical, molecular and in vivo data in one panel
Champions integrates clinical annotation, molecular characterization and in vivo pharmacology data to assemble panels around a specific program question. For KRAS, that includes whole-cell proteomics, phosphoproteomics and cell surface proteomics measured from tumor material in profiled models, because RNA expression does not reliably predict protein expression or pathway activation.
01
Clinically annotated models
Low-passage PDX models with clinical annotation, about half from metastatic, pretreated patients, so model selection can reflect real treatment history.
02
Prior KRAS benchmarking
We tested adagrasib across patient-derived xenograft models before its 2022 FDA approval, so KRAS benchmarking is not new for our team.
03
Published PDX concordance
Published work reports 85% positive predictive value for clinical disease control and 91% negative predictive value for progression (Izumchenko et al., 2017).
FAQ
What teams ask before choosing a panel
Answers on scope, flexibility, and what a panel study can and cannot tell you.
It depends on how many distinct subgroups your question needs to separate. A single-allele, single-indication question needs fewer models than one comparing alleles or tumor types. A KRAS scientist will recommend a size once the question is defined, and the starting sizes can be adjusted.
No. A panel shows how your compound behaves against a benchmark in patient-derived models and generates mechanism, biomarker, and combination hypotheses. It does not predict clinical response or guarantee an outcome, and it should inform, not replace, your own study design decisions.
Final model lists are confirmed with you during scoping, against current availability and external shareability. We share the candidate list and the rationale for each model before you commit, so you can review the selection with your own team.
Yes, where it changes your next decision. Endpoint tissue collection and deeper omic analysis can be scoped as additions to any panel, along with comparator arms or allele-specific inhibitor arms. We recommend adding them only when they are tied to a question your program has to answer.
A KRAS scientist joins the first scientific discussion, and scientific support continues through model selection, study design, and readout interpretation where scoped. Our bioinformatics and computational biology team supports data analysis when the scope includes it.
Resources
KRAS reading and posters
Posters, webinars, and articles from our KRAS team on pan-RAS(ON) response and the preclinical questions that follow.
Related services
Services that pair with a panel
A panel study produces tumor material and response data that feed directly into these services, from model sourcing to downstream omics and tissue analysis.
KRAS preclinical platform
The full KRAS offering, including the biology and evidence behind these panels.
Solid tumor PDX
The clinically annotated PDX inventory that panel models are drawn from.
Next generation sequencing
Whole exome and RNA sequencing behind each model's molecular profile.
4D proteomics
Whole-cell and phosphoproteomic measurement for follow-up on signaling hypotheses from panel tumor material.
Start a panel
Tell us the allele, indication, and question
Include the KRAS allele and tumor type you care about, how daraxonrasib fits your question, if at all, and the decision the study has to support. A KRAS scientist will review it before we schedule a model-selection discussion.
A KRAS scientist reads your request
We propose a panel design and size
No commitment until you approve a scope