Biology

Theoretical computer science

Bipartite graph

Cancer research

Biochemistry

Data mining

Computer science

Potency

Drug

Computational biology

Chronic myeloid leukaemia

Graph

In vitro

Drug response

Pharmacology

Artificial intelligence

Cluster analysis

Medicine

Myeloid leukaemia

Nature Communications

Combination therapy is preferred over single-targeted monotherapies for cancer treatment due to its efficiency and safety. However, identifying effective drug combinations costs time and resources. We propose a method for identifying potential drug combinations by bipartite network modelling of patient-related drug response data, specifically the Beat AML dataset. The median of cell viability is used as a drug potency measurement to reconstruct a weighted bipartite network, model drug-biological sample interactions, and find the clusters of nodes inside two projected networks. Then, the clustering results are leveraged to discover effective multi-targeted drug combinations, which are also supported by more evidence using GDSC and ALMANAC databases. The potency and synergy levels of selective drug combinations are corroborated against monotherapy in three cell lines for acute myeloid leukaemia in vitro. In this study, we introduce a nominal data mining approach to improving acute myeloid leukaemia treatment through combinatorial therapy.

Bipartite network models to design combination therapies in acute myeloid leukaemia