The phenomenon termed ”3D-domain swapping” occurs when two or more identical protein subunits relocate their secondary structure components, like strands or helices, to generate a different tertiary structure. This results in the creation of new protein-protein interactions and the formation of higher-order structures. This paper proposed an unsupervised computation approach for prediction of 3D-domain swapped cases by using various medicinal plant based protein sequence. Preparation of suitable data set is one of the crucial step for which Principal component attribute transformer (PCAT) and Percentile split (PS) have been used as a data mining technique followed by various attribute selection methods. To prepare the data set 11 impactful new attributes have been augmented in previously selected 39 attributes. To identify the groups of 3D-domain swapped protein with similar attributes an unsupervised algorithm modified K-Mean was implemented. In which curated 81 protein sequence with 4050 instances have been used and achieved 3 clusters for top ranked attributes. To measure the effectiveness of the proposed method Model-driven multivariate control chart (MDMCC) was applied with out of control point 3 along with the Within-cluster sum of square error (WCSSE) which ranges from minimum parameter value 0 to maximum parameter value up to 1.25. In-depth examination of generated clusters allows identifying the plants with desirable medicinal qualities and accelerates the decision-making process for possible plant candidates for pharmaceutical development. It also leads to the discovery of functionally related genes or biological pathways.