Fig. 1

MFMD pipeline. (1) In Preprocessing step MFMD uses DUST to remove sub-sequences with low complexity entropy. If DUST can not be run, MFMD uses an objective function defined in [21] to mitigate this problem. (2) In Pattern Discovery step, MFMD attempts to find the best PSSM matrix using GRASP and VNS heuristics. (3) In Pattern Matching step, MFMD uses the PSSM matrix found in the previous step to predict the initial positions of the motifs in the dataset