SARS-CoV-2 is a highly infectious virus responsible for the COVID − 19 pandemic. The increased transmission rate led to the emergence of SARS-CoV-2 variants. In viral infection, the receptor-binding domain (RBD) proteins are essential role in binding to the host receptor. Others, Heparan sulfate (HS), widely distributed on the surface of host cells, is thought to play an important role in the viral infection cycle of SARS-CoV-2. Therefore, it might be a reasonable strategy for antiviral drug design to interference with the RBD in the HS binding site. In this study, we used computational approaches to analyze multiple sequences of coronaviruses and reveal important information about the binding of HS to RBD in the SARS-CoV-2 spike protein. Our results showed that the potential hot-spots, including F456, R457 and S459 in RBD, exhibited strong interactions in the HS-RBD binding region. Therefore, we screened different compounds in the natural product database towards these hot-spots to find potential antiviral candidates using LibDock and MD simulation in Discovery Studio 2019. The results showed six potential natural compounds, including acetoside, chrysin 6-C-arabinoside 8-C-glucoside (CAG), hyperoside, isoquercitrin, oroxyloside and chrysin 6-C-glucoside 8-C-arabinoside (CGA) had strong binding ability to the RBD. Our results demonstrate a feasible approach to identify potential antiviral agents through the evaluation of the binding interaction between viral glycoproteins and host receptors. The present study provided the applications of the structure-based computational approach for the design and development of new antiviral drugs against SARS-CoV-2 variants.