Skip to main content
. Author manuscript; available in PMC: 2014 Jan 13.
Published in final edited form as: Inf inference. 2012 Dec;1(1):21. doi: 10.1093/imaiai/ias002

Table 9.

Denoising distances at various levels of noise for the ubiqutin (PDG 1d3z, [19]). η̄SDP,GOOD denotes the average noise of the hard constraint (‘good’) distances after the SDP embeddings. Note that the current implementation of the SNL-SDP algorithm used for the patch embeddings does not allow for hard distance constraints

Existing distances
Missing distances
η (%)
ANE¯old
ANE¯new
η̄ η̄SDP η̄SDP,GOOD η̄MDA η̄SDP η̄MDA
0 0.00048 0.00039 0 0.02 0.14 0 0.38 0.32
10 0.00316 0.00266 4.91 3.42 1.79 3.13 3.2 2.99
20 0.00590 0.00506 9.81 7.01 3.68 6.44 6.18 5.81
30 0.00957 0.00803 14.74 10.75 5.59 9.88 9.51 8.95
40 0.01292 0.01120 19.65 14.5 7.62 13.42 13.55 12.8
50 0.01639 0.01462 24.57 18.1 9.82 16.93 18.24 17.35