Timezone: »

 
Poster
PEER: A Comprehensive and Multi-Task Benchmark for Protein Sequence Understanding
Minghao Xu · Zuobai Zhang · Jiarui Lu · Zhaocheng Zhu · Yangtian Zhang · Ma Chang · Runcheng Liu · Jian Tang

Thu Dec 01 09:00 AM -- 11:00 AM (PST) @ Hall J #1032
in Poster Session 5 »

We are now witnessing significant progress of deep learning methods in a variety of tasks (or datasets) of proteins. However, there is a lack of a standard benchmark to evaluate the performance of different methods, which hinders the progress of deep learning in this field. In this paper, we propose such a benchmark called PEER, a comprehensive and multi-task benchmark for Protein sEquence undERstanding. PEER provides a set of diverse protein understanding tasks including protein function prediction, protein localization prediction, protein structure prediction, protein-protein interaction prediction, and protein-ligand interaction prediction. We evaluate different types of sequence-based methods for each task including traditional feature engineering approaches, different sequence encoding methods as well as large-scale pre-trained protein language models. In addition, we also investigate the performance of these methods under the multi-task learning setting. Experimental results show that large-scale pre-trained protein language models achieve the best performance for most individual tasks, and jointly training multiple tasks further boosts the performance. The datasets and source codes of this benchmark will be open-sourced soon.

Keywords: Multi-task Learning · Protein Modeling Benchmark · Protein Sequence Understanding
[ Paper]  [ Slides]  [ Poster]  [ OpenReview] 

Author Information

Minghao Xu (Montreal Institute for Learning Algorithms, University of Montreal, University of Montreal)
Minghao Xu

I was born in Shanghai, a fast-developing metropolis in China. Currently, I am a first year PhD student at Mila - Quebec AI Institute, advised by Prof. Jian Tang, where I focus on molecular and protein representation learning for drug discovery, and structural image representation learning for visual understanding.

Zuobai Zhang (Montreal Institute for Learning Algorithms, University of Montreal, University of Montreal)
Jiarui Lu (Mila - Quebec AI Institute)
Zhaocheng Zhu (Mila - Quebec AI Institute)
Yangtian Zhang (Montreal Institute for Learning Algorithms, University of Montreal, Université de Montréal)
Ma Chang (University of Hong Kong)
Ma Chang

I am a Ph.D. student at The University of Hong Kong, Department of Computer Science, co-advised by Dr.Lingpeng Kong and Dr.Tao Yu. My main research interest is in representation learning, along with interests in natural language processing and computational biology. I am passionate about developing new computational methods to study applicational problems, as well as improving the generalization ability of deep learning.

Runcheng Liu (Carnegie Mellon University)
Jian Tang (Mila)

More from the Same Authors