Hongzhi Wen
 |
Hongzhi Wen 文鸿志 CV Github Linkedin |
Short Bio
I am the Cofounder and Chief AI Scientist at CartaBio Inc, where I lead a team of scientists in applying advanced artificial intelligence to accelerate drug discovery. My professional journey began with a Ph.D. in Computer Science, where I focused on the intersection of artificial intelligence, deep learning and complex biological data. This led me to my current role, where my company builds sophisticated AI platforms to solve critical challenges in drug discovery. We develop foundational models for biology, analyze spatial omics data to find new biomarkers, and predict cellular responses to novel therapeutics.
Before founding CartaBio, I am a Ph.D. graduated from DSE lab at Michigan State University. My advisor is Professor Jiliang Tang.
I received my Bachelor's degree from Peking University in 2021, and used to work with Prof. Weiwei Sun. I am very grateful for the time I spent there.
More previously, I was an intern in Didi Global company from Oct
2019 to Mar 2020, Songguo company from
Oct 2020 to July 2021, Tigergraph in Summer 2022 and Amazon in Summer 2023. I spent really nice time in those amazing companies, especially enjoying the startup atmosphere.
My research focus is to apply graph neural networks and transformers to large-scale real-world datasets, especially single-cell analysis. I'm also very interested in graph transformers, and have been maintaining a paper list related to it.
Research Interests
Foundation Models for Scientific Research
Graph Transformers
Real-world Applications of Deep Learning
Selected Publications
CellPLM: Pre-training of Cell Language Model Beyond Single Cells. ICLR 2024. link
Single-Cell Multimodal Prediction via Transformers. CIKM 2023. link
Graph neural networks for multimodal single-cell data integration. SIGKDD 2022. link
DANCE: A Deep Learning Library and Benchmark for Single-Cell Analysis. Bioarxiv 2022. link
TA Experiences
CSE 482, Fall 2022, website
News
2024.07, officially join CartaBio Inc as a Co-founder and Head of AI!
2024.05, Investigating OOD of GNNs got accepted by SIGKDD 2024! Another LLM alignment ITERALIGN got accepted by NAACL. Thanks to the co-authors!
2024.01, CellPLM got accepted by ICLR 2024! Another GNN+LLM paper got accepted at ICLR 2024, thanks to the co-authors!
2023.12, we release a foundation model, CellPLM, for scRNA-seq and spatial transcriptomics. The performance is amazing! Check this awesome blog and our github for more details.
2023.09, we release a new dataset, Amazon-M2 for text-based session recommendation at NeurIPS 2023 Datasets and Benchmark!
2023.08, one paper got accepted by CIKM 2023!
2023.06, joining Amazon as an applied science intern in Amazon Search Team at Palo Alto!
2023.04, two new preprints about transformers in single-cell analysis are now available on arxiv! paper1 paper2
2022.12, presenting our works in single-cell competition and OGB-LSC competition at NeurIPS!
2022.12, two posters got accepted by NeurIPS 2022 AI4Science workshop. See you in New Orleans!
2022.11, one abstract got accpeted by MLCB 2022 conference. I am glad to be chosen as one of the 16 oral presenters in the conference!
2022.11, our team DANCE ranked top 2% (24 out of 1220) and got a Silver Medal in a Kaggle single-cell competition.
2022.11, our team DSE-node won the second place in OGB-LSC competition MAG240M track. Here is the official leaderboard!
2022.08, our DANCE package (a deep learning tool for single-cell analysis) is pre-released! We are also maintaining some wonderful paper lists [1, 2], feel free to star them!
2022.07, one poster got accepted by a computational biology workshop. We are presenting it at Berkeley!
2022.05, one paper got accepted by KDD 2022!
2021.11, Our team DANCE won first place in overall ranking of Predict Modality track in the NeurIPS competition - Multimodal Single-Cell Data Integration!