Sheng Wang


Assistant Professor

Paul G. Allen School of Computer Science & Engineering

University of Washington, Seattle

Email: swang [at] cs (dot) washington (dot) edu
Students and Visitors

My current focus is NLP applications (e.g., word sequence -> protein sequence, text generation) and ML applications (e.g., graph representation learning, zero-shot learning) in biomedicine
I am looking for PhD students starting from Fall 2021.
I am happy to host (remote) undergradaute / graduate visitors in Summer 2021.
If you want to work with me, please feel free to send me an email with your CV.


About me

I pursue open-world biomedical discovery. Just like how the player can explore the virtual world freely in modern video games, I use my research to help scientists conduct biomedical discovery freely in the open-world setting (e.g., the new sample could belong to a class we have never seen before). This line of research leads to important biomedical discovery, such as predicting cells of a novel cell type and annotating functions of a novel gene set. I developed computational approaches to advance representation, classification and interpretation in never-before-seen biomedicine.

My research has been used in major biomedical institutes, including Chan Zuckerberg Biohub, NIH Center of Excellence for Big Data Computing, Mayo Clinic, NIH National Center for Advancing Translational Sciences and Stanford School of Medicine. My current passion is about investigating the genetic factors of a rare birth defect phenylketonuria.

Research Interests (Open-world Biomedicine)

My research interests toward open-world biomedical discovery can be summarized in these three specific areas:

Computational Biomedicine:
Single Cell Analysis, Systems Biology, Drug Discovery, Cancer Genomics.

Limited-data machine learning:
Zero-shot Learning, Meta learning, Transfer learning

Natural Language Processing Applications to Biology and Medicine:
Knowledge Graph, Neural Text Generation, Relation extraction.

Education and experience

Preprints (*-equal contribution, #-corresponding author)