Revise Chai's code for NF, incorporating NetMHCpan

[evanbiederstedt]

Related to the following: https://hub.docker.com/r/cmopipeline/netmhc

https://github.com/mskcc/vaporware/pull/314

[evanbiederstedt]

Working on this now

[evanbiederstedt]

https://github.com/taylor-lab/neoantigen-dev (edited)

#####
# Neoantigen prediction pipeline. Four main steps: 
#       (1) Genotype HLA using POLYSOLVER, 
#       (2) Constructed mutated peptide sequences from HGVSp/HGVSc 
#       (3) Run NetMHC-4.0 + NetMHCpan-4.0
#       (4) Gather results and generate: 
#               - <sample_id>.neoantigens.maf: original maf with neoantigen prediction columns added
#               - <sample_id>.all_neoantigen_predictions.txt: all the predictions made for all peptides by both the algorithms
#####

Here’s what I would do here:

(i) Allan already has Polysolver running. (Killed it, boom). (ii) Take the Polysolver outputs and parse them into Python, that’s Chai’s (2) (iii) Feed into NetMHC (iv) use Chai’s python code to create outputs we need for science

I would recommend re-writing how he is parsing the config---make these individual args via argparse

[evanbiederstedt]

This is the script without the config parsing:

https://github.com/taylor-lab/neoantigen-dev/blob/feature/work_with_Nexflow/neoantigen.py

This should only need the CDS and cDNA references.

I didn't require arguments for the following:

    ## netmhc4_bin = config.get('NetMHCpan', 'netmhc4_bin_path')
    ## netmhc4_alleleslist = config.get('NetMHCpan', 'netmhc4_alleleslist')
    ## netmhcpan4_bin = config.get('NetMHCpan', 'netmhcpan4_bin_path')

In the current docker image cmopipeline/netmhc,

• netmhc4_bin is /data/netMHC-4.0/Linux_x86_64
• netmhc4_alleleslist is /data/netMHC-4.0/data/allelelist
• netmhcpan4_bin_path is /data/netMHCpan-4.0

These variable names are used in the script though---we should probably change this

[evanbiederstedt]

Allan killed it, boom https://github.com/mskcc/vaporware/pull/344