TL;DR
A beta 2 x 2 generated-pose triage for isolated arms in dual-target CAR concepts.
Use it / Skip it
It creates intended-pair and counter-screen Interface poses, then summarizes model-triage flags and artifacts without claiming binding or specificity.
Use when
You have two binder arms and two antigen ECD sequences and need early structural triage before wet-lab or construct work.
Don't use for
Do not use it to establish binding, specificity, or whether a fused binder-linker-binder construct folds correctly. It is isolated-pair pose triage only.
Inputs
- Binder arms
- Two antibody/VHH/scFv binding-arm sequences.
- Antigens
- Two antigen ECD sequences.
- Context
- Optional linker/name notes for record keeping.
Outputs
- 2 x 2 matrix
- Two intended-pair poses plus two alternate-pose counter-screens.
- Child jobs
- Links to Interface jobs with PDB and analysis CSV artifacts.
- Verdict/report
- Overall verdict, matrix CSV, and review notes.
Walkthrough
1. Paste two binding arms and two antigen ECDs
Paste two binding arms and two antigen ECDs.
2. Optionally load the dry-run example to understand the workflow, then replace with real sequences
Optionally load the dry-run example to understand the workflow, then replace with real sequences.
3. Run assessment
Run assessment. The parent job creates four child Interface jobs.
4. Inspect suspicious rows by opening the child Interface jobs
Inspect suspicious rows by opening the child Interface jobs.
5. Record review feedback before using the result in construct decisions
Record review feedback before using the result in construct decisions.
Under the hood
- Parent assessment is persisted separately from child Interface jobs.
- Each matrix cell reuses Interface prediction, contact extraction, BSA, pDockQ, PDB, and CSV artifacts.
- The page is beta and intentionally keeps standard Interface unchanged.
Worked example
Pitfalls
- This is not fused-construct folding.
- A clean matrix is not functional CAR validation.
- Weak antigen sequence quality creates misleading child jobs.