By: Michelle Vu
DrugMint derived the criteria for a drug-like molecule from the similarities among experimental and marketed drugs in the DrugBank database. For example, this analysis suggested that druglike compounds contain less primary alcohol, phosphoric monoester, diester and mixed anhydride, when compared to experimental compounds. DrugMint’s most groundbreaking quality is that it was engineered using compounds that are accessible to the public, from DrugBank, and that it has posted its open-source software on a web server. This access allows researchers unprecedented access to chemo-analytical software, which was only available commericially, to improve their experiments.
Explore DrugMint yourself by using its sample molecule! The webserver’s tools guide researchers through the initial stages of drug development. Users can first draw a molecule using the Marvin applet, which creates its molecular fingerprint, the digital data of a compound’s structure. Then, they can upload this molecule and run DrugMint’s virtual screening software, which assesses the drug-likeness of the molecule. If the assessment is unfavorable, users can modify the molecule’s structure using a Design Analogs Module. In addition, the website includes specialized modules for drugs designed to target cancer cells, HIV, and other notable diseases.
Some scientists are skeptical about the DrugMint’s usefulness, including Dr. Robert Murphy. Criticizing its accuracy and computational design he stated, “There is a potentially serious concern with the validity of the results due to the fact that the experimental design may result in overfitting.” However, DrugMint provides a powerful tool in initially screening for the best drug candidates; it provides a preliminary hypothesis and direction for drug development. Until we have developed a better knowledge base for human biochemistry, DrugMint, and other computational servers, will catalyze the rate of drug discovery.