I am a Computational Biology Ph.D. student at Tsinghua University, advised by Prof. Qiangfeng Cliff Zhang.
I am broadly interested in deep learning, with data-driven algorithm design and applications in real world (2D/3D computer vision and graphics) and micro world (Electron Cryo-Microscopy Density map based Molecular Structure Determination, Single Cell Clustering and RNA Structure Understanding). My recent research focuses on designing algorithms and deep learning models to gain biological insights from advances in the
large data sets.
Prior to Ph.D. student period, I was a visiting scholar at the Chinese University of Hong Kong, collaborating with awesome folks in the lab of
Xiaogang Wang. I also works closely with Zhe Wang,
Jianping Shi and Hongsheng Li, focusing on 3D Object Detection and
Molecular structure model building.
I received master and bachelor degree in Computer Science in Tianjin Polytechnic University in 2016 and 2013, respectively. I have also spent time at Multimedia Information Processing Lab,
Peking University as a research intern.
Feel free to say hi: xuk16 at mails.tsinghua.edu.cn
[Jan 2021] One paper accepted to Science. Proud to apply our improved A2-Net to identify new proteins in Spliceosome,
congratulations to Rui Bai and Ruixue Wan.
[Jan 2021] Our PrismNet method paper accepted to Cell Research, congratulations to all the team members.
[Nov 2020] Our PrismNet Prediction on SARS-CoV-2 paper accepted to Cell, congratulations to Lei Sun, Panpan and Wenze.
[Oct 2020] One paper accepted to Nucleic Acids Research, congratulations to Panpan and Xiaolin.
[Jul 2020] One paper accepted to Bioinformatics.
[Jul 2020] Proud to apply our PrismNet to predict host cell target proteins bind to SARS-CoV-2 RNA! Paper out on BioRxiv.
[May 2020] PrismNet method paper out on BioRxiv: Predicting dynamic cellular protein-RNA interactions using deep learning and in vivo RNA structure,
code are available at Github.
[Jan 2020] One paper accepted to ICRA 2020.
[Oct 2019] Paper out on Nature Communications: SCALE method for single-cell ATAC-seq analysis via latent feature extraction, code are available at SCALE.
[Mar 2019] Paper out on BioRxiv: VRmol: an Integrative Cloud-Based Virtual Reality System to Explore Macromolecular Structure, code are available
[Jan 2019] Presented our work A2-Net at AAAI 2019, Hawaii.
In vivo structural characterization of the whole SARS-CoV-2 RNA genome identifies host cell target proteins vulnerable to re-purposed drugs
Predicting dynamic cellular protein-RNA interactions using deep learning and in vivo RNA structure
Structure of the activated human minor spliceosome
RASP: an atlas of transcriptome-wide RNA secondary structure probing data
VRmol: an Integrative Web-Based Virtual Reality System to Explore Macromolecular Structure
SCALE method for single-cell ATAC-seq analysis via latent feature extraction
A2-Net: Molecular Structure Estimation from Cryo-EM Density Volumes
SegVoxelNet: Exploring Semantic Context and Depth-aware Features for 3D Vehicle Detection from Point Cloud
Bioactive functionalized monolayer graphene for high-resolution cryo-electron microscopy
RISE: a database of RNA interactome from sequencing experiments
Reweighted sparse subspace clustering
, a deep learning tool that integrates experimental in vivo RNA structure data and RBP binding data for matched cells to accurately predict dynamic
RBP binding in various cellular conditions.
provides the visualization and analysis of macromolecule structures in an infinite virtual environment on the web. VRmol is natively built with
WebXR technology, providing functions in a fully immersive, inspiring virtual environment.
Kitti Object Visualization
Data Transformation and Visualization for Kitti Object Detection Dataset
, including mapping 3D boxes on LiDar point cloud in
volumetric mode, mapping 2D and 3D boxes on Camera image, mapping 2D boxes on LiDar Birdview and mapping LiDar data on Camera image.
3D DenseNet (torch)
Implementing a 3D DenseNet
for 3D CAD Object Classification on ModelNet
dataset using highly
. VoxNet, C3D, and 3D-VGG models are also provided. Multi-GPU training is well supported. Check ModelNet Benchmark Leaderboard
to see related researches.
3D Deep Learning
A collaborative list of 3D Deep Learning works via 3D Representation, 3D Classification, 3D Classification, 3D Object Detection, 3D Reconstruction & Generation, 3D Human Pose Estimation and their related datasets.
A collaborative list of awesome CryoEM (Electron Cryo-Microscopy) resources, including Methods and Softwares related Particle Picking, 3D Classification, Denoising, Motion Correction, 3D Reconstruction, Model Building and some other related research.