A maximum likelihood framework for protein design

BMC Bioinformatics

Table 1 Specificity of designed sequences.

Potential	Average Z-score ratio	SDev Z-score ratio	Ranking (median)	Target fold in top 1% (A)	Target fold in top 10%	Average seq. identity	Correlation between (A) and mean entropy/site
μ	-0.12	0.18	2249	0.5%	4.8%	5.76 %	-0.26
μ + α_14ac	-0.10	0.18	2090	0.4%	6.3%	6.65 %	-0.04
μ + ε	0.13	0.16	816.8	10.7%	33.5%	6.69 %	0.23
μ + α_14ac + ε	0.45	0.23	32.7	53.6%	77.5%	7.82 %	0.64

Scores of a fold recognition experiment for designed sequences (see Methods). 1,200 sequences were sampled from p(s | c, θ) for each potential, and submitted to THREADER for fold recognition. Z-score ratio: Z-score of designed sequence/Z-score of native sequence in target fold.

ISSN: 1471-2105