“…Recently, computational and machine learning approaches for the sequence-based prediction of paratopes (Deac et al, 2018;Kunik et al, 2012b;Liberis et al, 2018), epitopes (Kringelum et al, 2012) or paratope-epitope (antibody-antigen) interaction (Baran et al, 2017;Deac et al, 2018;Jespersen et al, 2019;Kilambi and Gray, 2017;Krawczyk et al, 2013) are accumulating (for a more complete list of references see here: (Brown et al, 2019;EL-Manzalawy et al, 2017;Esmaielbeiki et al, 2016;Raybould et al, 2019a)). While the accuracy for the sequence-based prediction of paratopes seems generally higher than that for epitopes, overall, prediction accuracy has remained suboptimal -especially with regard to sequence-based prediction of epitope sites and paratope-epitope interaction (Brown et al, 2019;Greenbaum et al, 2007).…”