Ph.D. Candidate in Bioinformatics at Peking University | AI for Biology & AI4Science

I develop deep learning models that connect genomic sequence, gene regulation, spatial context, cell–cell communication, and cellular perturbation.

This research pipeline summarizes the main trajectory of my work, from long-range genomic sequence modeling and gene regulation to spatial context, cell–cell communication, and perturbation-oriented virtual cell modeling.

My research focuses on building interpretable and multimodal AI systems for understanding how genomic information shapes cellular states and tissue organization.

Long-range genomic sequence modeling for regulatory evolution

An interpretable deep learning framework for modeling ~180-kb genomic regulatory sequences and identifying cis-regulatory changes associated with cross-species gene expression evolution.

Role: First author · Lead developer
Status: Under review at Nature Structural & Molecular Biology
Code: Official Code

Interpretable spatial modeling of cell–cell communication

An interpretable multi-view graph neural network framework for identifying niche-specific cell–cell communication from spatial transcriptomic data.

Role: First author · Lead developer
Publication: Nature Computational Science (2025)
Links: Paper · Official Code

Multimodal modeling of genomic regulation and spatial cellular context

A multimodal AI framework integrating genomic regulatory sequence, cell identity, spatial microenvironment, and cell–cell communication for spatial gene expression modeling and in silico perturbation.

Role: First author · Lead developer
Status: Manuscript in preparation
Code: Official Code

Inferring candidate perturbations from target cellular states

A computational framework for identifying candidate upstream perturbations that may drive cells toward desired molecular states, with the goal of enabling inverse design of cellular phenotypes.

Role: Project lead · Lead developer
Status: Ongoing research
Code: Official Code

Single-cell spatial mapping and spatial multi-omics reconstruction

A computational and web-based framework for mapping single cells to spatial locations using spatial transcriptomics references and extending spatial information to multiple molecular layers.

Role: Co-first author
Publication: Nucleic Acids Research (2023)
Links: Paper · Web Server

• Juntian Qi†, Zhengchao Luo†, et al.
  Interpretable niche-based cell–cell communication inference using multi-view graph neural networks
  Nature Computational Science, 2025.
  First author · Lead developer

• Juntian Qi, Shuhan Yang†, et al.
  DeepEvo deciphers the cis-regulatory grammar of human evolution to prioritize adaptive and disease drivers
  Nature Structural & Molecular Biology, under review.
  First author · Lead developer

• Juntian Qi, et al.
  GenoSpatial integrates genomic sequence and spatial context to predict context-dependent gene expression
  Manuscript in preparation.
  First author · Lead developer

• Xiangshang Li†, Chunfu Xiao†, Juntian Qi†, et al.
  STellaris: a web server for accurate spatial mapping of single cells based on spatial transcriptomics data
  Nucleic Acids Research, 2023.
  Co-first author

† Equal contribution.

• AI & Deep Learning: long-range genomic sequence modeling, multimodal learning, graph neural networks, hypergraph neural networks, contrastive learning, attention mechanisms, and representation learning
• AI for Biology: gene regulation modeling, spatial microenvironment modeling, cell–cell communication, virtual perturbation, and inverse design of cellular states
• Computational Biology: single-cell RNA-seq, spatial transcriptomics, regulatory genomics, epigenomics, and cross-species transcriptomic analysis
• Model Interpretation: in silico sequence perturbation, regulatory variant interpretation, motif analysis, feature attribution, and biological hypothesis generation
• Research Engineering: Python, PyTorch, R, Scanpy/AnnData, HDF5, Linux, Git, and reproducible model training pipelines

• Email: juntian_qi@163.com
• Google Scholar: Juntian Qi