Cellpeutics.偲礼生物
EN中文

AI-ASSISTED TCR DISCOVERY & COMPUTATIONAL ASSESSMENT

From tumor targets to TCR candidates for validation

Connect target evidence, AI design and structural screening to develop evidence-informed candidates for joint experiments.

Discuss collaboration Explore the workflow
A green TCR approaches a coral target peptide presented by blue HLA between two cell membranes.
  • TCR
  • Target peptide
  • HLA
AI-generated concept illustration, not an actual candidate structure or experimental image.

What does a TCR recognize?

Cells present protein fragments on HLA molecules. A TCR recognizes this combination.

What are we looking for?

Traceable research candidates with computational support that warrant further experimental assessment.

HOW THE PLATFORM WORKS

A visual path to research candidates

Define the target, then propose candidates. Prescreen sequences and public patent information before structural calculations, and carry the evidence into subsequent validation.

Define the target, establish peptide-HLA context, propose candidates, prescreen sequences and public patents, compare structures and organize a research panel; joint experimental validation follows.
Different stages have been implemented across project phases. Supporting evidence and model disagreements are retained to inform subsequent experimental design.

Expand a step to explore its research question.

  1. 01 / Define the target

    Establish the target, peptide and HLA context.

  2. 02 / Establish recognition context

    Check presentation evidence and peptide-HLA structures.

  3. 03 / Propose TCR candidates

    Explore paired sequences within research constraints.

  4. 04 / Prescreen sequences and patents

    Check sequence validity and public patent neighbors before structural comparisons.

  5. 05 / Compare multiple structural models

    Compare contacts, clashes and disagreements between models.

  6. 06 / Build a research panel

    Organize candidates, evidence and questions for subsequent experiments.

Explore the research methods and tools

Tools used at different stages include ProteinMPNN, TCRDiff, tFold-TCR, TCRmodel2, MHCflurry, PMGen and PRODIGY. Platform work focuses on workflow integration, rule management and evidence traceability. Tool and data use follows the applicable licenses.

COMPUTATIONAL RESEARCH TO DATE

From a method benchmark to different targets

Method benchmark

MAGE-A4

Use public experimental structures for candidate design and comparisons across multiple models.

Cross-target research

WT1

Screen candidates, separate reference information and compare after freezing the research outputs.

Mutant-epitope exploration

PIK3CA

Prioritize patent prescreening, then assess candidate structures and diversity.

Application exploration

KK-LC-1 (CT83)

Organize a small research candidate panel for discussion of subsequent validation.

FINDING RESEARCH PARTNERS

Start with what you already have

An existing target, TCR or experimental validation capability can provide the starting point for a clearly scoped, small research question.

A tumor target, candidate TCR or experimental resources feed into a jointly scoped study, candidate evidence, validation questions and subsequent experimental plans.
Illustrative collaboration directions. Research scope, responsibilities and rights are agreed separately.

FINDING RESEARCH PARTNERS

Let’s discuss a research question

What is your target? What data do you have? What would you like experiments to answer? We welcome research discussions; scope, responsibilities and rights are agreed separately.

Discuss your research needs

A research overview is enough for initial discussions. Share confidential materials through agreed channels after appropriate arrangements are in place.

Back to TCR services in AIDD