# Shanghai AI Lab — “长江杯”全球AI蛋白质设计挑战赛赛题公布，上海AI实验室邀全球研究者共解序列谜题

- Company: Shanghai AI Lab (shlab.org.cn)
- Announced: 2026-11-05
- Category: not stated
- Coverage: not counted
- Announcement: no
- Group: routine
- Source: https://www.shlab.org.cn/news/5444315
- Record: https://forck.live/items/33293-ai-ai
- Title in English (translated by forck.live from zh): "Yangtze River Cup" Global AI Protein Design Challenge problems announced, Shanghai AI Laboratory invites global researchers to solve sequence puzzles together
- Subject: InternLM / 书生

Shanghai AI Laboratory has published the seven target tracks and full competition rules for the "Yangtze River Cup" 2026 Global AI Protein Design Challenge, which it launched on September 24 with more than ten partner institutions. The organizers say 1,800-plus teams and over 4,000 participants from more than 20 countries and regions have registered so far. The challenge plans free wet-lab validation for roughly 7,000 designed sequences, and the lab's Intern-Discovery scientific discovery platform is the official recommended platform, offering competitors full-stack research support and free computing power during the contest. Each of the seven tracks accepts only one specified molecule type, and teams may enter one or several tracks, with screening, validation and ranking done separately per track. The minibinder track requires 40-100 amino acid single polypeptide chains, while the VHH track requires 100-150 amino acid single VHH variable domains, both built only from the 20 standard natural amino acids and submitted as mature protein bodies without signal peptides, tags or Fc fusions. Sequences must be designed from scratch and checked against a reference library by CDR region, and registration and sequence submission close on November 30.

## Evidence

Verbatim from https://www.shlab.org.cn/news/5444315:

> On September 24, the Shanghai AI Laboratory, together with more than ten leading institutions, launched the 'Changjiang Cup' 2026 Global AI Protein Design Challenge. As of now, over 1,800 teams and more than 4,000 participants from over 20 countries and regions have registered for the competition. Recently, the challenge announced the targets for seven tracks and the complete competition rules, and plans to provide free wet-lab validation for approximately 7,000 designed sequences. The Intern·Duanyan Scientific Discovery Platform, independently developed by the Shanghai AI Laboratory, serves as the official recommended platform for this competition, offering participants full-stack research support and free computing power during the event. Researchers in AI and life sciences from around the world are invited to participate with their models and designs, bringing AI-designed proteins into reality. Registration and sequence submission deadline: November 30. Changjiang Cup Challenge official website: https://aiprotein.sjtu.edu.cn/ Intern·Duanyan experience link: https://discovery.intern-ai.org.cn/chat Seven targets, seven tracks: This competition sets up seven independent tracks, each accepting only one specified molecular type. Each team may participate in one or more tracks. Each track is independently screened, validated, and ranked for awards. *Target-related information is based on sequence (UniProt ID) and sequence range; reference structures are for display only. Competition design requirements: [minibinder track] Amino acid length: 40-100 aa; Molecular form: single polypeptide chain, single designed binding module; Sequence requirements: 20 standard natural amino acids; submit only the mature protein body, without signal peptides, tags, Fc, or other fusion segments. [VHH track] Amino acid length: 100-150 aa; Molecular form: single VHH variable domain, including complete framework and CDRs; Sequence requirements: 20 standard natural amino acids; submit only the mature protein body, without signal peptides, tags, Fc, or other fusion segments. *Submitted sequences must be de novo designed; sequences must be aligned with the CDR region of a reference library. Sequence submission method: 1. Go to the competition website and log in using the [team leader email + team leader phone number] submitted during registration. Username: team leader email; initial password: team leader phone number (*Please change your password after logging in). 2. Enter the submission portal and select the corresponding track. 3. Upload submission materials: Material 1: Designed sequence (*must be in CSV format as required); Material 2: Design summary (*disclosure of model weights, training set, or source code is not required). 4. Confirm submission. More information.

---

Record: https://forck.live/items/33293-ai-ai
Catalogue: https://forck.live/llms.txt
Current issue: https://forck.live/feed.md
