Fig. 2

Methodology diagram. In the example, the method receives raw sequences containing proteins 1, 2 and 3 as input. Once aligned, it is possible to segment each protein. Then, the normalization and tokenization protein sequence processes are performed. Subsequently, amino acid co-occurrence matrix sets are generated for each protein, which will be classified by an individual RF for each protein. Finally, each RF is interpreted by Shap Values, thus generating explanations for each protein