ProcessNets: Towards an efficient approach for ensemble analysis of biological networks

In the realm of systems biology, examining the characteristics of multiple networks related to a biological system is crucial. However, handling thousands of large networks presents a significant challenge. Traditional methods like parallelism and optimized algorithms, which address this issue by treating each network as an independent task, fall short in harnessing the similarities between networks originating from the same biological system.

To overcome this hurdle, researchers propose a novel phylogeny-like data structure called nc-tree, designed to compactly represent an ensemble of related biological networks. Additionally, they introduce a versatile framework named ProcessNets, which employs incremental algorithms over the nc-tree to expedite property computations on input networks.

Extensive analysis of space and time complexity, along with empirical evaluations utilizing diverse ensembles of simulated and real-world GTEx gene coexpression networks, reveal that ProcessNets offers substantial space and time advantages when computing various network science measures on ensembles of similar networks. For example, a compression factor of 2.1x and 8.1x is achieved in storing two representative ensembles of GTEx Muscle Skeletal networks containing 1000 nodes/genes, while the speedup in computing degree centrality ranges from 2.86x to 3.97x over the fastest baseline for ensembles with at least 14,500 nodes derived from subsampled GTEx datasets of ten tissues.

These findings hold promise and suggest the potential application of ProcessNets for analyzing biological network ensembles derived from rapidly accumulating consortium/biobank-based datasets.