Skip to content

đź“ť [Paper] TCRen: predicting TCR-peptide recognition based on residue-level pairwise statistical potential

License

Notifications You must be signed in to change notification settings

antigenomics/tcren-ms

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

33 Commits
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

TCRen pipeline is free for acadimc and non-commercial use. Inquiries regarding commercial use can be e-mailed to the last author (PI) of the study.

TCRen

TCRen is a method for prediction of TCR recognition of unseen epitopes based on residue-level pairwise statistical potential

TCRen method starts from a structure of the peptide-MHC complex with the TCR of interest—either experimentally derived or based on a homology model—then extracts a TCR-peptide contact map and estimates the TCR-peptide energy of interaction for all candidate epitopes using TCRen potential, which we derived from statistical analysis of amino acid contact preferences in TCR-peptide-MHC crystal structures.

preview

Dependencies

R with packages data.table, tidyverse, optparse, stringr, magrittr (tested on: R v4.0.5 with data.data.table v1.14.0, tidyverse v.1.3.1, optparse v1.7.3, stringr v1.4.0, magrittr v2.0.3; R v4.2.0 with data.data.table v1.13.0, tidyverse v.1.3.0, optparse v1.7.3, stringr v1.4.0, magrittr v1.5; we expect other package versions should also be compatible with the script)

Java (tested on openjdk 11.0.16)

Repository content

  • Script to run TCRen pipeline on new target TCRs and candidate epitopes. This script is provided in 2 versions: 1) as Rmarkdown file TCRen_pipeline/run_TCRen.Rmd which can be run in Rstudio (the first chunk should contain paths to input files); 2) as R script TCRen_pipeline/run_TCRen.R which can be run from a command line (with arguments indicating paths to input files; for details see Tutorial below)

  • Example files for input and output (folder example)

  • Scripts and data to reproduce the benchmarking of TCRen performance and other analysis performed in the corresponding paper (scripts in the folder code_paper, data in the folder data)

  • All cleaned-up TCR-peptide-MHC structures from PDB (folder data/PDB_structures) with meta-data (data/summary_PDB_structures.csv) and the file with all extracted TCR-peptide residue contacts (data/contact_maps_PDB.csv)

  • Values of TCRen potential (TCRen_potential.csv)

TCRen input

  1. A structure of TCR-peptide-MHC complex (either experimentally derived or a homology model). Several structures may be submitted at once. Structure(s) should be placed in a single folder.
  • Example: example/input_structures
  1. A list of candidate epitopes.
  • Example: example/candidate_epitopes.txt

TCRen output

A table with 4 columns: complex.id (corresponding to the name of an input structure), peptide (corresponding to the name of a candidate peptide), potential (“TCRen” if the default TCRen_potential.csv file is used) and score (TCRen estimate of energy of peptide-TCR interaction).

  • Example: example/output_TCRen/candidate_epitopes_TCRen.csv

Tutorial

  1. Clone the github repository for TCRen:

$ git clone https://github.com/antigenomics/tcren-ms.git

  1. Prepare a structure (or several structures) of TCR-peptide-MHC complex. Format: “.pdb”.
  • All structures for which predictions will be done should be placed in a single directory (e.g. example/input_structures)

  • If for the TCR of interest a crystal structure of the ternary complex (TCR-peptide-MHC) with some peptide is available (i.g. for the task of prediction of cross-reactivity of a well-known TCR), it can be downloaded directly from PDB and used as input. For the task of predictions for unseen TCRs, homology model(s) should be used as input, e.g. obtained using TCRpMHCmodels tool which is implemented both as a webserver and a stand-alone software. Detailed instructions for TCRpMHCmodels tool use can be found on the webserver site and in README of software download. The modeling usually requires a few minutes for a single TCR-peptide-MHC complex either in the web-server or in a single CPU.

  1. Prepare a list of candidate epitopes (e.g. mutated peptides predicted as binders to host MHC for the task of prediction of neoepitope recognition). Format: “.txt” file with each candidate epitope in a separate line, with a header “peptide”. Note that only peptides with the same length as in the input structures would be considered for predictions.
  • Example in example/candidate_epitopes.csv
  1. Run TCRen pipeline (typical runtime is within minutes for a standard laptop).
  • The pipeline should be run the directory containing all the files necessary for TCRen launching: run_TCRen.R, TCRen_potential.csv and mir-1.0-SNAPSHOT.jar (java script for annotation and extraction of contacts in TCR-peptide-MHC structures)

  • Detailed comments on all stages of TCRen pipeline can be found in Rmarkdown file run_TCRen.Rmd which is identical in terms of the main code to run_TCRen.R

$ cd TCRen_pipeline

$ Rscript --vanilla run_TCRen.R -s ../example/input_structures/ -c ../example/candidate_epitopes.txt -p TCRen_potential.csv -o ../example/output_TCRen/ -m 1G

  • Arguments of run_TCRen.R script:
option Description
-s, --structures directory with input structures
-c, --candidates file with candidate epitopes
-p, --potential file with energy potential (TCRen)
-o, --out directory for TCRen output
-m, --memory memory allocation
  1. The output of TCRen can be found in the file candidate_epitopes_TCRen.csv in the folder which was set by -o flag. The content of the output file is described above in the section “TCRen output”

About

đź“ť [Paper] TCRen: predicting TCR-peptide recognition based on residue-level pairwise statistical potential

Resources

License

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published