AutoSite2- A fast and accurate protein-ligand binding-site prediction method



AutoSite2 identifies the most suitable pockets for a ligand to bind to a macromolecule. It proceeds in two steps. First, AutoSite2 analyzes the high affinity regions over the receptor and find pockets of varying sizes. In the second step, AutoSite2 selects and ranks pockets according to their suitability to hold a ligand with a given volume.

This tutorial is designed to you get started with the command line execution of the program to obtain binding sites represented as fill points, for a receptor (pdb:1wbz), First go to the Downloads page to obtain the software and install it following the installation instruction.


To run this tutorial you will also need a receptor file in PDBQT format (example) and potentially a ligand (example)

Step 1: Receptor analyze.

Linux command:


> $WHERE_YOU_INSTALLED/MGLTools2-latest/bin/pythonsh $WHERE_YOU_INSTALLED/MGLTools2-latest/MGLToolsPcks/AutoSite/bin/ -bc ratio 1.2 -sc -steps 4 -r 1wbz_P_recFixed.pdbqt -o ./


MacOSX command:


> /Library/MGLTools2/latest/bin/pythonsh /Library/MGLTools2/latest/MGLToolsPckgs/AutoSite/bin/ -bc ratio 1.2 -sc -steps 4 -r 1wbz_P_recFixed.pdbqt -o ./


-bc is a command line option allowing to specify the lower bounds of affinity cutoff values used by AutoSite2 when it scans affinity cutoff values. In this particular the lower bounds are given as a ratio of the default cutoff values used by AutoSite, i.e. -bc ratio 1.2 means to use 1.2 * default cutoff values.

-sc informs AutoSite to scan a range of cutoff values (i.e. perfrom the AutoSite Algorithm to search for pockets with various sizes).

-steps defines the number of steps used in AutoSite2 algorithm.


The result of this command line is a save in a file describing a tree structure that can be queried to identify the best pockets for a given ligand volume.


Step 2:  Pocket selection and re-ranking.


With a given ligand file.

Linux command:


> $WHERE_YOU_INSTALLED/MGLTools2-latest/bin/pythonsh $WHERE_YOU_INSTALLED/MGLTools2-latest/MGLToolsPcks/AutoSite/bin/ -f 1wbz_P_pep.pdb -p True


MacOSX command:


> /Library/MGLTools2/latest/bin/pythonsh /Library/MGLTools2/latest/MGLToolsPckgs/AutoSite/bin/ -f 1wbz_P_pep.pdb -p True


Or with a given volume:


Linux command:


> $WHERE_YOU_INSTALLED/MGLTools2-latest/bin/pythonsh $WHERE_YOU_INSTALLED/MGLTools2-latest/MGLToolsPcks/AutoSite/bin/ -s 900 -p True


MacOSX command:


> Library/MGLTools2/latest/bin/pythonsh /Library/MGLTools2/latest/MGLToolsPckgs/AutoSite/bin/ -s 900 -p True


AutoSite2 peptide scoring function is used with flag “-p True”



Step 3:  Results.


In the original folder where AutoSite2 is executed, all the pockets in the tree structure are saved and the topology information is saved in pockettree.txt. Pockets that best match the ligand volume are in the folder named “pockets” with their rankings indicated in the file name.