Shuaike Shen

Shuaike Shen

Zhejiang University

🧑🏻‍💻 About me

I am currently a senior undergraduate student from Mixed Class, Chu Kochen Honors College of Zhejiang University, majoring in Computer Science and Technology. My advisors are professor Chunhua Shen and Hao Chen who are affiliated to State Key Lab of CAD & CG, Zhejiang University. I had a wonderful research experience at University of Illinois at Urbana-Champaign this summer and was honored to work with Prof. Ge Liu

I am keenly interested in research topics associated with AI for Science and various Generative Models. My prior research experience encompasses protein design and the prediction of crystal structures and properties. Currently, I am engaged in exploring diverse areas, including co-designing protein structures and sequences, utilizing Reinforment Learning to guide Flow Matching models to design protein structures.

I am looking for PhD position for Fall 2025 intake, if you have any clues please feel free to contact me.

🔥 News

  • 2024.06:  🎉🎉 One paper has been accepted by ICML’2024 AI for Science Workshop.
  • 2024.05:  🎉🎉 One paper has been accepted by ICML’2024.
  • 2024.01:  🎉🎉 The first paper I participated in was accepted by ICLR’2024.

📝 Publications

ICML'2024
sym

Floating Anchor Diffusion Model for Multi-motif Scaffolding

Ke Liu*, Shuaike Shen*, Weian Mao*, Xiaoran Jiao, Zheng Sun, Hao Chen†, Chunhua Shen†

[Project Page] [Code] [Paper]

Introduction

  • A method to scaffold multiple protein functional motifs without prior knowledge. FADiff is based on a diffusion process which is modified to take into account the movement of the single motifs as independent rigid bodies.
ICML'2024 AI for Science Workshop
sym

Boost Your Crystal Model with Denoising Pre-training

Shuaike Shen*, Ke Liu*, Muzhi Zhu, Hao Chen†

[Project Page]

[Paper]

Introduction

  • A universal pre-training framework based on denoising can enhance the model’s ability to model crystal structures and can be migrated to any specific crystal feature extraction network.
ICLR'2024 spotlight
sym

De novo Protein Design Using Geometric Vector Field Networks

Weian Mao*, Zheng Sun*, Muzhi Zhu*, Shuaike Shen, Lin Yuanbo Wu, Hao Chen†, Chunhua Shen†

[Code] [Project Page] [Paper]

  • The code will be released after being sorted out.

Introduction

  • We propose a novel structure graph encoder, which can address the atom representation bottleneck observed in traditional IPA encoders.

🌐 Research Interests

AI for Science

  • Protein & Antibody Design

  • AI for Drug Discovery

  • Materials discovery & properties prediction

Generative Models

  • Diffusion models

  • Flow Matching

  • Equivariant Neural Networks

🏯 Research Experience

Computer Science & Biochemistry, University of Illinois at Urbana-Champaign (UIUC)

Advisors: Prof. Ge Liu & Prof. Nicholas Ching Hai Wu (May.2024 - Present)

  • Protein sequence and structure co-design.

  • Apply Reinforcement Learning to guide Flow Matching models to design protein structures.

State Key Lab of CAD & CG, Zhejiang University

Advisors: Prof. Chunhua Shen & Prof. Hao Chen (Oct.2022 - May.2024)

  • Contributed to the VFN project, addressing the atom representation bottleneck observed in traditional IPA encoders.

  • Fully participated in the FADiff project and solved the multi motif scaffolding problem.

  • Proposed a Pre-training framework and Period Injection Module that can enhance crystal modeling capabilities

📖 Educations

  • 2021.09 - 2025.6 (now), Zhejiang University, Zhejiang
  • 2018.09 - 2021.06, Urumqi No.1 Senior High School, Xinjiang