. 2017 Dec 27;13(12):e1005905. doi: 10.1371/journal.pcbi.1005905

Table 1. Benchmark of peptide-protein complexes used in this study (non-redundant set; see S1C Table for full set).

PDB ids of the initial calibration set are highlighted in bold.

PDB ID	Peptide sequence^a/ secondary structure	Fragment similarity^b	PIPER^c	PIPER-FPD (motif)					PIPER-FPD (full peptide)
Complex/Free Receptor	Peptide sequence^a/ secondary structure	Fragment similarity^b	L^d	L		I^e		Fnat	L	I		Fnat
Known binding motif (n = 12)
`1CZY:CE` `1CA4:A`	`PQQATDD` `CEECCCC`	2.2(2.5)	17.6	1.6	0.6		0.86		2.4		0.9		0.76
`1EG4:AP` `1EG3:A`	`NMTPYRSPPPYVP` `TTTTTTCCCCCCC`	0.6(3.0)	21.6	12.9	4.1		0.19		29.3		11.5		0.00
`1ELW:AC` `1A17:A`	`GPTIEEVD` `CCCCCCCC`	1.0(2.4)	3.2	0.8	2.5		0.75		2.6		3.0		0.71
`1JD5:AB` `1JD4:A`	`AIAYFIPD` `CEEEETCC`	0.7(2.3)	2.8	1.2	0.5		0.88		8.2		2.9		0.19
`1JWG:BD` `1JWF:A`	`DEDLLHI` `CCCCCCC`	2.8(2.9)	3.4	2.2	0.8		0.90		2.2		0.8		0.90
`1MFG:AB` `2H3L:A`	`EYLGLDVPV` `CCCCCCEEC`	1.5(2.7)	3.1	1.8	0.8		0.73		8.7		3.1		0.29
`1NTV:AB` `1P3R:B`	`NFDNPVYRKT` `CEETTTTCCC`	2.7(3.3)	5.6	5.2	1.7		0.43		3.8^f		1.5		0.52
`1RXZ:AB` `1RWZ:A`	`KSTQATLERWF` `CEEECTTTTTC`	2.4(4.0)	7.4	5.0	1.9		0.31		3.2		1.7		0.39
`1SSH:AB` `1OOT:A`	`GPPPAMPARPT` `CCCCCCCCCCC`	1.1(2.3)	2.6	7.6	2.9		0.56		1.9^f		1.1		0.87
`1X2R:AB` `1X2J:A`	`LDEETGEFL` `CTTTTTCCC`	0.2(0.5)	1.1	1.3	0.5		0.74		1.7		0.6		0.72
`2A3I:AB` `2AA2:A`	`QQKSLLQQLLTE` `CCCCHHHHHHHC`	0.3(3.5)	1.6	1.0	0.4		0.93		4.8		2.1		0.72
`2CCH:DF` `1H1R:B`	`HTLKGRRLVFDN` `TTTTCCCCCCCC`	1.8(4.6)	3.4	1.0	0.4		0.91		3.9		1.6		0.67
No known binding motif (n = 15)
`1AWR:CI` `2ALF:A`	`HAGPIA` `CCCCCC`	1.6	6.5						1.3		0.5		0.97
`1ER8:EI` `4PAE:A`	`HPFHLLVY` `CCCBCCBC`	1.9	4.1						1.2		0.8		0.80
`1LVM:AE` `1LVB:B`	`ENLYFQ` `CCEEEC`	1.9	2.3						1.4		0.6		0.91
`1NVR:AB` `2QHN:A`	`ASVSA` `CEEEC`	2.0	5.5						7.1		2.2		0.56
`1NX1:AC` `1ALV:A`	`DAIDALSSDFT` `HHHHHHHHHCC`	1.8	1.9						1.3		0.9		0.80
`1OU8:BD` `1OU9:A`	`GAANDENY` `CCCCCCCC`	3.1	6.1						6.4		2.4		0.39
`1U00:AP` `2V7Y:A`	`ELPPVKIHC` `CCCCEECCC`	2.3	6.0						2.1		1.7		0.71
`2B9H:AC` `2B9F:A`	`RRNLKGLNLNLH` `CCTTTTCCCCCC`	3.2	16.4						15.8		6.3		0.18
`2C3I:BA` `2J2I:B`	`KRRRHPSG` `CCCCCCCC`	2.5	3.5						8.7		2.5		0.26
`2DS8:BP` `2DS7:A`	`ALRVVK` `CCEECC`	1.4	4.7						1.2		0.6		0.86
`2FMF:AB` `1JBE:A`	`QDQVDDLLDSLGF` `HHHHHHHHHHHCC`	1.1	5.2						1.3		0.7		1.00
`2H9M:CD` `2H14:A`	`ARTKQ` `TTTTC`	2.2	4.7						4.0		1.1		0.37
`2HPL:AB` `2HPJ:A`	`DDLYG` `CCCCC`	1.6	3.0						1.4		0.5		0.91
`2O02:AP` `2BQ0:A`	`GLLDALDLAS` `THHHHHCCCC`	2.4	2.8						3.4		1.2		0.74
`3D1E:AP` `3D1G:A`	`GQLGLF` `CBCCCC`	2.5	4.1						6.7		2.0		0.51

^a Motif (as defined by the ELM database [29]) is underlined (motif details are provided in the S1B Table).

^b Similarity between fragments and bound peptide conformation: Median backbone RMSD (Å) (in parentheses: results for full peptide).

^c Results for PIPER simulations are given for the motif / full peptide for known/unknown motifs, respectively. The models are selected as in PeptiDock [24] (without the minimization step).

^d,e Modeling accuracy: L–Ligand RMSD (models within 2.5 Å are highlighted in bold)^c; I–Interface RMSD ^d. Defined as in the CAPRI experiment [34, 35].

^f Complexes for which docking of the full peptide provides better models than docking of the motif only.