Name
Abstract | ||
|---|---|---|
Identifying drug-target interactions is an important step in drug discovery, but only a small part of the interactions have been validated, and the experimental determination process is both expensive and time-consuming. Therefore, there is a strong demand to develop the computational methods, which can predict potential drug-target interactions to guide the experimental verification. In this paper, we propose a novel algorithm for drug-target interaction prediction, named Neighborhood Constraint Matrix Completion (NCMC). Different from previous methods, for existing drug-target interaction network, we exploit the low rank property of its adjacency matrix to predict new interactions. Moreover, with the rarity of known entries, we introduce the similarity information of drugs/targets, and propose the neighborhood constraint to regularize the unknown cases. Furthermore, we formulate the whole task into a convex optimization problem and solve it by a fast proximal gradient descent framework, which can quickly converge to a global optimal solution. Finally, we extensively evaluated our method on four real datasets, and NCMC demonstrated its effectiveness compared with the other five state-of-the-art approaches. |
Year | DOI | Venue |
|---|---|---|
2018 | 10.1007/978-3-319-93034-3_28 | ADVANCES IN KNOWLEDGE DISCOVERY AND DATA MINING, PAKDD 2018, PT I |
Keywords | Field | DocType |
Low rank matrix completion,Neighborhood constraint,Drug-target interaction | Adjacency matrix,Data mining,Mathematical optimization,Drug discovery,Gradient descent,Computer science,Drug target,Exploit,Interaction network,Convex optimization,Constraint matrix | Conference |
Volume | ISSN | Citations |
10937 | 0302-9743 | 0 |
PageRank | References | Authors |
0.34 | 15 | 5 |
Name | Order | Citations | PageRank |
|---|---|---|---|
| Xin Fan | 1 | 776 | 104.55 |
| YuXiang Hong | 2 | 2 | 1.71 |
| Xiaohu Liu | 3 | 1 | 1.03 |
| Yaogong Zhang | 4 | 2 | 1.73 |
| Maoqiang Xie | 5 | 36 | 7.62 |