highly significantp -value < 0.001
p -values were FDR-corrected with the Benjamini–Hochberg method and only highly significant correlations ( p -value < 0.001), supported by at least three of the four similarity measures, were kept: this procedure compensates the method biases [ 37 ].