These file was made from a library of 240 structures at 1.7A resolution or better. Care was taken to remove side chains which had uncertain positions (eg high B-factors) or systematically misfit conformations. A paper (see below) describes our methodology, and the advantages of this library over others previously published.
The approach taken in preparing this file differs from that of the standard O rotamer file. Firstly, although the most common rotamer comes first, subsequent rotamers are arranged so that neighbours have similar conformations. This allows you to narrow down your choice to a few rotamers which will be next to each other as you turn the dial.
Secondly, it contains many more rotamers. The original O library used only those rotamers with more than 5% occurrence in order to avoid artifacts (although there still were a few). In contrast, we have carefully tested all rotamers and distributions to avoid (we hope) any artifacts. Thus some rare-but-real conformations are included. This means that non- rotameric conformations should be viewed with even more suspicion than before, and never used in disordered surface positions.
Thirdly, it contains some non-rotamers! These are sample points in allowed, well populated, relaxed conformations. They are not, however, local energy minima and thus, strictly, not rotamers. For example, Glu chi3 has a very flat distribution which is populated throughout, so we have included the rotamer at chi3=0 and two sample points at chi3=60 and chi3=-60. There are sample points only for Glu, Gln, Asp, Asn, Phe and Tyr. They all have their frequency listed as 0%.
For more information see: