Figure 2

The flowchart of MKA showing the stepwise procedure. Figure 1 shows the stepwise procedure we have performed. (1) data collection, building Benchmark_3 and pre-processing datasets; (2) profile construction, such as PSSM, Sarah1 and EH-profile; (3) the information obtained in (2) and (3) were combined and normalised to fall in the interval [--1, 1] to be fed into networks; (4) target levels were assigned to each profile (positive, +1, for domain boundary residues and negative, --1, for non-boundary residues); (5) a hold-out method, to divide the combined dataset into seven subsets (training and testing sets); (6) model training on each set, to create a model; (7) simulation of each model on the test set, to obtain predicted outputs; and (8) post-processing to find predicted domain boundary locations. The procedure from (6) to (8) was performed iteratively until we obtained the most suitable kernel and the optimal hyperparameters for HME for Benchmark_3 dataset.