Table 1.

Summary of the popular protein–peptide complexes datasets that are widely used for testing and benchmarking different docking tool

Dataset	Number of complexes	Length of peptide	Special Features	Specific application	Availability
LEADS-PEP	53	3–12 residues	Diverse sequence of peptides, complexes do not interact with nucleic acids	Due to smaller peptide size, suitable for testing tools adapted from small molecule docking tools	www.leads-x.org
PeptiDB	105	5–15 residues	Diverse secondary structure of peptides including conformational change upon binding, complexes with diverse biological functions	Suitable for testing tools that tackle peptide flexibility	RCSB code of the complexes: https://ars.els-cdn.com/content/image/1-s2.0-S096921260900478X-mmc1.pdf
PPDbench	133	9–15 residues	Diverse in term peptide sequences (<40% sequence similarity) and biological functionalities	Suitable for testing docking tools on different complexes categorised with different functionalities	https://webs.iiitd.edu.in/raghava/ppdbench/
PepPro	89	5–30 residues	Contains 58 unbound receptors structures	Useful for testing tools whether they can predict apo-holo conformational change	http://zoulab.dalton.missouri.edu/PepPro_benchmark
Propedia	~20000	2–50 residues	Contains subsets of complexes based on clustering on different features such as sequence, interface structure or binding site	Broader range of peptide length allows it to test different type of docking tools. Also, different subset gives flexibility to user on testing their tools	https://bioinfo.dcc.ufmg.br/propedia
PixelDB	1966	NA	Uses machine learning to identify protein and peptide. This helps to overcome the issue of incorrectly identifying them when peptide is larger than the receptor	Broader range of peptide length allows any docking tools to be tested on	https://github.com/KeatingLab/PixelDB