This repository contains a python script for scoring mmps and nmers for their likelihood of recognition against a list of Class I alleles based upon the methods described in
"Development of a model for ranking candidate HLA class I neoantigens based upon datasets of known neoepitopes" Gartner et al.
This code is written to run with python version 3.6 and requires IEDB Immunogenicity tool, MHCflurry1.6, netMHCpan-2.8 and netMHCStabpan1.0. Links are provided below please accept user license and install on your system.
Two conda environment .yml's are distributed to create the python environment mhcflurry-env to run model. If for any reason they don't work instructions for creating the environment are in the instructions.txt file
IEDB immunogenicity prediction Tool (http://tools.iedb.org/immunogenicity/download/)
NetMHCStabpan 1.0 (http://www.cbs.dtu.dk/services/NetMHCstabpan/)
NetMHCpan-2.8 (http://www.cbs.dtu.dk/services/NetMHCpan/)
After cloning repository on system please read instructions full step by step description for setting up model found in instructions.txt. IEDB mmunogenicity needs to be converted to python 3, Instructions for doing so are in this file. After everything is installed alter IEDB immunogenicity and NetMHCStab paths in src/prediction_modules.py to their locations on your system.
Clone repository
git clone https://github.com/JaredJGartner/SB_neoantigen_Models.gitchange directory and unzip MmpModel
cd SB_neoantigen_Models/src/models/
gunzip mmp_Model.pickle.gz
gunzip nmer_Model.pickle.gzUsing your favorite text editor edit the paths to IEDB imuunogenicity, netmhcstabpan to their location on your system and save changes.
A full step by step description of setting model is found in instructions.txt
Example input file is provided in examples so that you can test that everything is working properly
1) An excel sheet with the following header:
Unique identifier,Wt nmer,Mut nmer,Exome VAF decile,Gene expression decile,Present in RNA-seq data
Unique identifier = a unique ID to link all data back to; Wt nmer = Wild type amino acid sequence; Mut nmer = Mutant amino acid sequence; Exome VAF decile = The variant allele frequency for all mutations broken into decile for more information see "Development of a model for ranking candidate HLA class I neoantigens based upon datasets of known neoepitopes" Gartner et al.; Gene expression decile = The expression of the gene binned into deciles as based upon al genes expression for that sample; Present in RNA-seq data = 1 or 0 for whether mutation was found in RNA sequencing data 1 = yes, 0 = no.
2) a list of up to 6 Class I HLAs formatted to match netMHC input for example HLA-A01:01
*3) optionally you can provide the flag --cpus and increase the number of cpus this will speed up the neMHCstab prediction portion by spreading the commands across cpus
Test to see outputs are the same as those given in examples
ensure that conda environment is Active. directions to create environment are in instructions.txt
conda activate mhcflurry-envmkdir test cd test python ../src/GenerateScores.py ../examples/nmer_test_input.xlsx HLA-A02:01 HLA-A03:01 HLA-B13:02 HLA-B15:01 HLA-C05:01 HLA-C06:02
This will produce a single output excel files in this test directory same as seen in examples directory
## License
This project is licensed under the GNU General Public License v3.0 - see the [LICENSE.md](LICENSE.md) file for details