CompareCBS

Structural comparison of binding sites based on atom types and distances

Prerequisites

To use CompareCBS, you need to have R installed along with the following packages:

• igraph
• bio3d
• cowplot
• fastcluster
• openxlsx
• vanddraabe

Recommended Installation using Mamba

To set up the required environment, follow these steps:

1. Create a new environment and install the necessary R packages:

   mamba create -n legacy_pepit r-base r-essentials r-bio3d r-igraph r-cowplot r-fastcluster r-openxlsx
   mamba activate legacy_pepit

2. Install the vanddraabe package manually, as it has been removed from the CRAN repository:

   wget https://cran.r-project.org/src/contrib/Archive/vanddraabe/vanddraabe_1.0.0.tar.gz
   R

3. In the R console, install the downloaded package:

   install.packages('vanddraabe_1.0.0.tar.gz')

Pepit Installation

To install the legacy pepit package, open R in the folder containing the package file and run:

install.packages('pepit_1.0.tar.gz')

File Format

Binding site files should be formatted as classical PDB files with one modification: atom typing should be specified in column 77 (1-based index).

DIONYSUS Atom Typing

Protein:

• Cα: A
• Aromatic carbon: a
• Cβ: b
• Sidechain oxygen: o
• Backbone carbon: C
• Backbone oxygen: O
• Backbone nitrogen: N
• Sidechain carbon (non-aromatic): c

Carbohydrate:

• C1: s
• Other carbons: g
• Oxygens: u

Different chains can be selected during the comparison process. It is advisable to rename chains according to the precise selections you intend to compare.

Binding Site Comparison

To compare binding sites, use the following command:

Rscript sugar.R [target_pdb] [chains] [query_pdb] [prefix]

Example

Rscript sugar.R data/5EYX_1_MAN_1.pdb A,B data/1DGL_1_MAN_3.pdb examples/jacalin_vs_l-type

Output Files

The comparison script outputs four types of files:

• .al: Alignment file containing the correspondence between each mapped atom.
• .index: Summary of the .score file.
• .score: Contains alignment metrics, notably the coverage (number of atoms mapped), alignment mean distortion (mean dist), and its score (coverage weighted by atomic pair distortion).
• .pdb: Contains the aligned query to the target.

By default, if the same prefix is used multiple times, sugar.R will append the .al, .index, and .score files and create new .pdb files.

References

[1] Rasolohery, Inès, Gautier Moroy, and Frédéric Guyon. "PatchSearch: a fast computational method for off-target detection." J Chem. Info. Model. 2017.
[2] Rey, Julien, Inès Rasolohery, Pierre Tufféry, Frédéric Guyon, and Gautier Moroy. "PatchSearch: a web server for off-target protein identification." Nucleic Acids Res, 2019.
[3] Guyon, Frédéric, and Gautier Moroy. "Non-sequential alignment of binding sites for fast peptide screening." bioRxiv, 2023.
[4] More recent version of pepit https://github.com/DSIMB/pepit