Fig 4. Complementary Rotamers Improve Binding Site Sequence Recovery.

A. The fraction of designable positions in the benchmark complex set where the correct (wild-type) residue identity was incorporated at least 50% of the time (native sequence recovery) for a range of special_rot bonus values. Control indicates native sequence recovery when using the unmodified Packer. B. Box and whisker plots for profile similarities at designable positions in the benchmark set, where designable positions that achieve a profile similarity of >0.9 regardless of modifications to the Packer have been removed (plots with all positions are shown in S2 Fig). Box and whiskers denote quartiles, where whiskers indicate minimum and maximum values, boxes indicate 25^th and 75^th percentiles, respectively, and the line indicates the median value. C. Scatter plots showing profile similarity for individual designable positions at various special_rot bonus values (left: 0; middle: -1.5; right: -4.0) compared to the unmodified Packer. D. Scatter plots showing shannon entropy for individual designable positions at various special_rot bonus values (left: 0; middle: -1.5; right: -4.0) compared to the unmodified Packer.