Dear Macromolecular structure community, I am a technical research assistant working with Prof. Joel Sussman, currently head of the PDB. I approach the question of formatting and organizing molecular coordinate information from the point of view of a molecular modeller and programmer, rather than as a database specialist. Compared with the new proposed mmCIF and ZINC standards, the existing classical PDB format is remarkably well suited to being read and written by modelling applications, as well as being conservative on storage and network transmission resources, owing to its compactness. Indeed, it has apparently become the accepted standard among the vast majority of modelling packages. While the so-called "HEADER" portions of PDB entries should rightfully be organized in a manner most conducive to on-line data management, search and retrievel, the so-called "ATOM" portions (eg. the ATOM, HETATM and TER records) should be retained in their present format, or something very similar. These latter record types both are much less likely to be the subjects of direct data-searches (Any searches concerning molecular geometry most likely would require some sort of intermediary software to derive it from the coordinates.) and constitute the bulk of most macromolecular data entries. However, some minor modifications might be considered, to allow atom names to five characters, residues to four characters, and the X-, Y- and Z-coordinates to six decimal places in place of the existing three. The less-important atom number and comment fields could be moved to the right, beyond column 80. Indeed, Prof. Sussman has remarked, and I agree, that, no matter what format is adopted, every given data item should be presented on a single line, even if that line extends beyond column 80. Nowadays, having longer line lengths does not present the kinds of problems it once did, and several Unix and ed tools such as 'sort' and 'grep' operate on the basis of single lines, and cannot handle multi-line formats. Even scripts and macros are much harder to write for multi-line formats. -- Sincerely yours, Clifford Felder <felder@sgjs2.weizmann.ac.il>