Table 2. Refinement, validation and analysis of the deposited models.

The table contains statistics of interest for model comparison in §§ 3.2–3.4. The PDB headers include the full information.

Model	N14C3, conservative	N14C3, optimistic	N14A3
Refinement
PDB entry	5l9d	5l88	5l7x (5lgh)
R free (5% set)	0.215	0.211	0.228
R work	0.176	0.170	0.190
ΔR	0.039	0.041	0.038
TLS groups	4 (VH, VL + glycan, CH1, CL)	4 (VH, VL + glycan, CH1, CL)	4 (VH, VL + glycan, CH1, CL)
No. of atoms
Protein	3280	3300	3224
‘Ligand’	97	189	64
Waters	221	252	223
All refined non-H	3598	3741	3511
〈B〉 (Å2)
Protein atoms	33.5	35.6	39.0
‘Ligand’ atoms	61.5	74.4	59.2
Waters	40.0	41.6	41.5
All refined non-H atoms	34.7	38.0	39.5
Refined occupancy groups	12	13	4
Glycans	Asn26L-NAG	Asn26L-NAG-FUC	Asn26L-NAG
PEG fragments	7	14	5
Missing residues	6	3	16
Coordinate errors (Å)
Free	0.139	0.140	0.139
σA	0.119	0.120	0.141
Cruickshank DPI	0.154	0.156	0.154
F o versus F c correlation	0.967	0.970	0.968
F o versus F c correlation, free	0.956	0.956	0.955
R.m.s.d., bond lengths (Å)	0.009	0.009	0.011
R.m.s.d., angles (°)	1.36	1.37	1.50
Geometry
Elbow angle (°)	139	139	147
Clashes, true, reported	4 of 10	5 of 11	3 of 3
Ramachandran†
Total	431	436	412
Allowed	7	8	6
Outliers	0	0	0
Real-space R outliers, RSRZ > 2	8	9	10
Side-chain rotamer outliers	6	9	14
Buried contact surface, reported, L+H‡ (Å2)	6350, 1834	9640, 1854	5250, 1700
LLDF ‘ligand’ outliers	5 of 9	10 of 17	5 of 6

^†As determined using the Ramachandran plot boundaries by Lovell et al. (2003 ▸).

^‡L+H signifies the actual surface contact area between protein residues of the L and H chains, obtained by excluding the solvent molecules from the contact calculation (cf. the output of the detailed contact tables provided by the PISA web service at http://www.ebi.ac.uk/pdbe/pisa/).