The site frequency spectrum (SFS) is a popular summary statistic of genomic data that records the number of mutations found at various frequencies in a population. For understanding genomic data from tumors, it is necessary to study how features of cancer, such as growth and selection, impact the shape of the SFS. Can the SFS at a given time point be utilized to understand properties of the cancer evolutionary process leading up to that time? We will explore these ideas using some branching process models of tumor growth.