01
RNA Intelligence
Developing generative models to understand and design RNA molecules by bridging sequence, structure, and biological constraints.
Focussequence · structure · topology
AI × RNA × Biomolecular Discovery
Zhiqi Ma
AI for Biomolecular Design & Scientific Discovery.
I develop generative and optimization methods for RNA, therapeutic sequences, and reliable molecular design — with the long-term goal of closing the loop between scientific reasoning, molecular generation, and biological experiments.
Visiting Student · MedAI Lab, Westlake UniversityJoint PhD Program Pre-admit · Zhongguancun AcademyGenerative AIRNA / siRNABiomolecular Design
Research
Three directions toward biomolecular discovery.
02
Therapeutic Biomolecular Design
Developing generative approaches for therapeutic molecules that bridge computational design and biological function, from RNAi candidates to antimicrobial peptides.
FocussiRNA · peptide · activity optimization
03
Reliable Biomolecular Generation
Developing evaluation frameworks that answer one question before anything reaches the lab: how do we know this candidate is reliable? Testing generative candidates under conflicting objectives and real-world trade-offs.
Focusevaluation · uncertainty · model–lab feedback
Publications & Preprints
Scientific questions behind the work.
siProGenA: Generative siRNA Candidate Construction via Position Proposal and Guide Generation Independently conceived
Scientific question: Where to target · What guide sequence to design.
Biological hypothesis: Can controlled deviations from canonical complementarity improve the potency–specificity trade-off in siRNA design?
Computational evidence: Across 354 measured sites, constrained variants outscored canonical complements under all four evaluators (85.1% on average). Matched-random, GC, and secondary-structure controls support the comparison.
siProGenA reframes siRNA design as generative Where/What candidate construction, producing compact transcript-level libraries intended to reduce synthesis and screening burden.
siRNA-mRNA Dual Diffusion Model for RNAi Drug Design Independently conceived
Scientific question: How to generate effective RNAi candidates with diffusion models under biological constraints.
A diffusion-based framework for RNAi candidate generation by modeling target context and guide sequence design.
PRAGA: Prototype-aware Graph Adaptive Aggregation for Spatial Multi-Modal Omics Analysis
Scientific question: How to learn robust biological representations from spatial multimodal omics data.
A prototype-aware graph learning framework for multimodal spatial omics representation.
Background
Experience & education.
Research Journey
Jul–Oct 2023
NetEase Games
Business Intelligence Researcher · Generative AI
Jun 2024–Jun 2025
Great Bay University
Research Assistant Intern · Prof. Xubin Zheng
Jun–Oct 2025
SJTU School of Medicine
Research Assistant Intern · Prof. Haicang Zhang
Jan–May 2026
Peking University
Research Assistant Intern
Why Biology?
I moved from general AI to biology because I wanted to work on questions where computational advances could deepen our understanding of living systems and ultimately create tangible value for human health.
Mathematics & ModelingUnderstanding complex systems through abstraction
Generative AILearning how AI can create and optimize
AI for BiologyMoving from medical imaging toward deeper biological questions
RNA & Therapeutic DesignConnecting sequence, structure, and function
Scientific DiscoveryLinking computation with experiments, translation, and real-world deploymentThe 23rd century will be the age of biology.
Education
M.Sc. in Bioinformatics
The Chinese University of Hong Kong, Shenzhen
2024–2026
B.Eng. in Information and Computing Science
Jinan University–University of Birmingham International Joint Programme
2020–2024
Selected Honors
National Second Prize
National College Students Mathematical Modeling Competition · Top 3% nationally
2021 & 2022
Academic Rising Stars Scholarship
Jinan University
2023
Long-term research vision
Beyond benchmark optimization.
UnderstandLearning representations that capture biological structure, function, and constraints.
Generate & ReasonBuilding AI systems that generate molecular designs and reason about scientific hypotheses.
ValidateTesting computational predictions through independent evaluation and biological experiments.
DiscoverTurning validated models and hypotheses into new scientific insights and real-world applications.
My long-term goal is to develop reliable AI systems that move beyond black-box optimization toward scientifically grounded discovery. I am particularly interested in connecting generative models, scientific reasoning, biological validation, and research translation—and in building scientific agents that can support this iterative process. As AI takes on more of the search, I want the questions to stay human — these tools should extend a researcher's judgment, not replace it.
About
I am a researcher working at the intersection of generative AI, computational biology, and scientific discovery.
I received my M.Sc. in Bioinformatics from The Chinese University of Hong Kong, Shenzhen, and I am currently a Visiting Student at Westlake University.
My research focuses on generative models for RNA and therapeutic biomolecules, particularly sequence–structure co-design, siRNA generation, and reliable molecular optimization. Across these problems, I am interested in how structural and biological constraints can be incorporated into generative models, and how model-guided optimization can be evaluated reliably.
Beyond Research
Outside research, I enjoy films, music, and a quiet drink with friends. One song I keep returning to is The Last Day of the End of the World by NoMBe.