With the increasing use of high-throughput sequencing technology in tumor research, a large number of somatic variations are being identified and some of them have proved to be responsible for tumorigenesis (Cancer Genome Atlas Research Network et al., 2013). Investigating structural and functional impacts of tumor somatic variants would greatly help to identify causal variations, understand the mechanisms of carcinogenesis, and develop novel anti-tumor therapies. Therefore, many efforts have recently been made to map genomic variations to 3D protein structure, such as G23D (Solomon et al., 2016) and G2S (Wang et al., 2018). Furthermore, Cancer 3D database (Porta-Pardo et al., 2015) and HotSpot3D (Niu et al., 2016) were developed to discover functional implications of mutations by means of structure data and drug information. However, there are still some limitations. Firstly, the effects of insertions and deletions (indels) are not taken into consideration. Secondly, these tools heavily depend on the resolved structures in Protein Data Bank (PDB) (Berman et al., 2000), i.e. they are not applicable when there is no reliable structural information available for wild-type protein. Here, we developed a webserver, TGPred, which provides a series of functionalities, including protein structure prediction, ligand binding site prediction, identification of functional relevant mutations, and estimation of functional impacts of mutations. Based on an interactive visualization design, these analyses are flexibly integrated, and thus the function impacts of a given protein variant could be inferred. The website is available at http://www.yyli-lab.cn/TGPred/.