Natural products, known as chemical compounds or substances produced by living organisms, are arguably the most important players in revolutionizing modern medicine. To date, natural products remain the best source of drug leads, including penicillin, streptomycin, artemisinin and others. However, one of the difficulties of natural product-based drug discovery is the structural derivatization. The traditional chemical modification methods often require multiple protection and deprotection steps for the large variety of functional groups, making the process laborious and problematic. A promising alternative approach to produce natural product derivatives is utilizing biosynthesis enzyme toolbox; for example, enzyme-based "glycol-randomization" and "alkyl-randomization" can dramatically expand diversities of important anti-cancer natural products. In my thesis work, I characterized seven biosynthesis enzymes (including sugar aminotransferase WecE, CalS13, AtmS13, glycosyltransferase SsfS6, OleD, methionine adenosyltransferase sMAT and methyltransferase DnrK) by X-ray crystallography. The structural characterizations and protein engineering of those enzymes lead to successfully expand the glyco/alkyl libraries, as well as to broaden the glyco/alkyl installation processes. These discoveries and enzyme toolbox developments are directly applicable to future drug discovery for cancer, and can be utilized as blue print to further understand the essential role of glycosylation and methylation in biology.