Ever since Dr. C. Anfinsen's pioneering work, it has been the fondest hope of many protein physical chemists to be able to predict the three-dimensional structure of a globular protein from its amino acid sequence. One of the major properties of the amino acids that are used towards this goal is their hydrophobicity. However, the extent to which the hydrophobicity properties of amino acid residues can aid in predicting the structure has not been quantitatively measured. In the past year, Dr. D. Lipman of the Arthritis Institute, Dr. R. Pastor of the FDA and I made a general study of the relationship between the hydrophobicity pattern and an important structural property--the accessibility of individual amino acid residues. Using crystal structures and the native and randomly shuffled sequence data sets, we found that we could indeed quantitatively measure the power of the hydrophobicity pattern in predicting the accessibility pattern of globular proteins. We have found, unfortunately, that there is a severe limitation in the amount of structural information one can extract from the hydrophobicity pattern represented by the native sequence data, implying that even rudimentary structure prediction must include considerations other than hydrophobicity alone.
