HXu @nyxflower

Heterogeneous graph representation learning aims to learn low-dimensional vector representations of different types of entities and relations to empower downstream tasks. Existing popular methods either capture semantic relationships but indirectly leverage node/edge attributes in a complex way, or leverage node/edge attributes directly without taking semantic relationships into account. When involving multiple convolution operations, they also have poor scalability. To overcome these limitations, we proposes a flexible and efficient Graph information propagation Network (GripNet) framework.

Hao Xu, Shengqi Sang, Peizhen Bai, Ruike Li, Laurence Yang, and Haiping Lu. "GripNet: Graph Information Propagation on Supergraph for Heterogeneous Graphs." Pattern Recognition (2023).

Leishmaniasis is an infectious disease with very few known effective medications. In this project, we proposed a computational pipeline to find new therapeutic targets for this disease, by combining multiple state-of-the-art machine learning and simulation methods: DL-based DeepPurpose, protein surface based MONN and docking simulation. Our method identified several potential drug candidates, whose efficacy was evaluated in the Zindi-Indaba competition.

Jonathan Smith, Hao Xu, Xinran Li, Laurence Yang, and Jahir Gutierrez. "Compound Screening with Deep Learning for Neglected Diseases: Leishmaniasis." Machine Learning in Computational Biology (2022).

For problems like drug side effects, a "predictor" that correctly predicts side effects is far from enough. A question of greater concern is: Can the "predictor" help us to figure out the cause of a predicted side effect? In this project we provided a positive answer to this question by introducing APRILE, a graph learning XAI framework for discovering the mechanism behind drug side effects.

Hao Xu, Shengqi Sang, Herbert Yao, Alexandra Herghelegiu, Haiping Lu, and Laurence Yang. "APRILE: Exploring the Molecular Mechanisms of Drug Side Effects with Explainable Graph Neural Networks." bioRxiv (2021).

Taking multiple medications at the same time can cause serious side effects. Therefore, we developed TIP, a biomedical knowledge graph based side effect predictor for the use of drug combinations. Actually, TIP model is an efficient general approach for multi-relational link prediction in any multi-modal (i.e. heterogeneous and multi-relational) network with two types of nodes. It can also be applied to other knowledge graph completion and recommendation tasks.

Hao Xu, Shengqi Sang, and Haiping Lu. "Tri-graph information propagation for polypharmacy side effect prediction." NeurIPS Workshop on Graph Representation Learning (2020).