This repository contains different network integration methods that can be classified into:
Unweighted approaches These methods perform network integration without considering the “predictiveness” (i.e. the informativeness of each network) with respect to a prediction task. In particular, the following integrations are implemented: - Unweighted Average (UA) - Per-edge Unweighted Average (PUA) - Maximum (MAX) - Minimum (MIN) - At least K (ATLEASTK)
Weighted approaches These methods require to provide as input a weight for each network, which are usually learned considering an appropriate learning algorithm: - Weighted Average Per-class (WAP) - Weighted Average (WA)
Citation - These methods were presented in the paper:
Valentini, Giorgio, et al. “An extensive analysis of disease-gene associations using network integration and fast kernel-based gene prioritization methods.” Artificial Intelligence in Medicine 61.2 (2014): 63-78.
Corresponding bib entry:
@article{valentini2014extensive,
title={An extensive analysis of disease-gene associations using network integration and fast kernel-based gene prioritization methods},
author={Valentini, Giorgio and Paccanaro, Alberto and Caniza, Horacio and Romero, Alfonso E and Re, Matteo},
journal={Artificial Intelligence in Medicine},
volume={61},
number={2},
pages={63--78},
year={2014},
publisher={Elsevier}
}