In prior study, our group have approved a tool for predicting Cross-interaction or/and Self-interaction of monoclonal antibodies using sequences, which called CISI.The the result of CISI is determined by the p-values of all the two models and CISI achieved an accuracy of 88.19%.

In this study, We used different feature extraction methods to characterize the sequencesand. The descending dimension method, MRMD2.0, based on the PageRank algorithm allowed us to get fewer features than CISI and the sensitivity of CISI2.0 achieved 100%.