Skip to main content
. 2017 Oct 25;27(1):146–158. doi: 10.1002/pro.3292

Table 4.

Accuracy of Predicting Coil 3JHN‐Hα Values for Disordered Peptides and Proteins

Protein N a RMS deviationsb [Hz]
<3J> 3Jna 3JANN <(3Jna+3JANN)/2> 3JSearle
Aβ(1–40) 30 0.45 0.38 0.37 0.37 0.42
α‐synuclein 111 0.39 0.29 0.29 0.29 0.50
Protease 64 0.44 0.38 0.42 0.38 0.43
Integrase‐N 44 0.47 0.41 0.36 0.37 0.58
Ubiquitinc 65 0.45 0.41 0.38 0.38 0.42
GGXGGd 18 0.43 0.43 0.28 0.31 0.37
All 332 0.43 0.37 0.36 0.35 0.47
a

Number of experimental 3J HN‐Hα couplings for each test set; for N‐terminally acetylated αsynuclein, the first six residues have partial helical character and were excluded; temperature correction48 was applied to experimental data if not measured at 20°C, including a +0.16 Hz correction for Aβ (measured at 4°C) and a −0.10 Hz correction for denatured ubiquitin data (measured at 30°C); pressure correction of +0.08 Hz/kbar49 is applied to the experimental data not acquired at 1 bar, including a +0.20 Hz correction for both pressure‐denatured HIV‐1 Protease and the N‐terminal DNA‐binding domain of HIV‐1 Integrase (both collected at 2500 bar).

b

RMSD between experimental and predicted 3J HN‐Hα coupling constants; <3J> is the difference relative to the average 3J HN–Hα value calculated using Eq. (2) for residues of type X in the newly generated coil library, 3Jna is calculated relative to <3J> plus the neighbor adjustments of Table 3; 3JANN is the RMSD relative to values predicted by the ANN‐based Eq. (4). 3JANN values for each of the 8000 triplets are also available at https://spin.niddk.nih.gov/bax/nmrserver/rc_3Jhnha/ann_results.txt; 3JSearle is the RMSD relative to the neighbor‐corrected coil values of Searle and co‐workers.15

c

From Peti et al.50

d

From Shi and Kallenbach.33