> For the complete documentation index, see [llms.txt](https://docs.hits.ai/hyperlab-release-note-en/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://docs.hits.ai/hyperlab-release-note-en/changelog_en/2026-07-29-relese-note.md).

# 2026-07-29 Relese Note

The update for HyperLab on July 29, 2026, has been completed.

#### HyperLab Version : 2026.07.29

***

## :mega: Changelogs

#### This update introduces a new **Cyclic Peptide Design** capability and the **LigandMPNN** model to the Bio Co-Scientist workspace.

You can now perform the entire cyclic peptide design process—from backbone generation and sequence design to structure prediction—within a single workflow. You can also design more refined protein sequences by considering the molecular environment of protein-bound partners, including small molecules, metal ions, DNA, and RNA.

### 1. New Cyclic Peptide Design Capability

<figure><img src="https://709489233-files.gitbook.io/~/files/v0/b/gitbook-x-prod.appspot.com/o/spaces%2FUd87mch9xATD0InicbJH%2Fuploads%2FqVEndd3gfoJYeLFgkaLu%2Fimage.png?alt=media&amp;token=d6580bd3-e586-442c-b7ee-41f3e733ca8a" alt=""><figcaption></figcaption></figure>

A new design pipeline has been added to automate cyclic peptide backbone generation, sequence design, and 3D structure prediction.

* RFdiffusion generates a cyclic peptide backbone structure.
* ProteinMPNN designs an amino acid sequence based on the generated backbone.
* The designed sequence is automatically converted into a cyclic peptide SMILES representation that reflects the cyclization bond between the N- and C-termini, which cannot be fully represented using a linear FASTA sequence alone.
* The converted SMILES is submitted to K-Fold to predict the final 3D structure.
* The required metadata is added to the predicted .cif file, allowing the structure to be displayed as a cartoon representation in the 3D Viewer.

The workflow can now proceed continuously through cyclic peptide structure prediction without requiring users to provide separate instructions for SMILES conversion.

### 2. New LigandMPNN Model

<figure><img src="https://709489233-files.gitbook.io/~/files/v0/b/gitbook-x-prod.appspot.com/o/spaces%2FUd87mch9xATD0InicbJH%2Fuploads%2FaNAHEIJnrOadJkF4eQhR%2F%E1%84%89%E1%85%B3%E1%84%8F%E1%85%B3%E1%84%85%E1%85%B5%E1%86%AB%E1%84%89%E1%85%A3%E1%86%BA%202026-07-29%20%E1%84%8B%E1%85%A9%E1%84%92%E1%85%AE%205.43.52.png?alt=media&amp;token=7ec959a9-a8ef-4e22-86ed-cdf05de37eee" alt=""><figcaption></figcaption></figure>

The LigandMPNN model has been added to support protein sequence design while considering the structural environment of molecules bound to the protein.

LigandMPNN can account for the local environment around binding partners such as:

* Small-molecule ligands, including drug-like molecules
* Metal ions
* DNA
* RNA

This enables protein sequence design that considers interactions with binding partners at ligand-binding pockets, enzyme active sites, metal-binding sites, and protein–nucleic acid interfaces.

In particular, this update improves the precision and applicability of sequence design for small molecule–protein, DNA–protein, and RNA–protein complexes.

***

## ⚒️ Bug Fixes

Bug fixes and service stability improvements have been applied.