New Scientist: A recent study applies the concept of pattern recognition used in computational linguistics to research on key molecular structures in organic chemistry. Bartosz Grzybowski of Northwestern University in Evanston, Illinois, and his colleagues treated molecules as sentences and molecular fragments as words. They then developed an algorithm to find fragments that are common among different molecules and rank their frequency. They also looked for the fragments with the highest information content. What they found was that the bonds connecting the most informative molecular fragments were often key bonds that connect simpler compounds and serve as building blocks for more complex molecules. Although the parallels between chemistry and language are not clear-cut, the tools the researchers have developed could prove useful in picking out unique chemical fragments that might be missed by human researchers and that could be used in the development of new compounds and pharmaceuticals.
