-
Notifications
You must be signed in to change notification settings - Fork 24
Command line interface
Installing the package provides the command epitopepredict in your path. This is a command line interface to the library without the need for any Python coding. It provides pre-defined functionality with settings specified in a text configuration file. Using this you can make MHC predictions with your chosen alleles and predictors. If you are using the IEDB prediction tools they should be installed locally and you can specify the path in the [iedbtools] section. Otherwise ignore those settings. Note that if settings are left out generally defaults will be used so you can have a minimal file as in the examples.
You can also choose to do additional analysis of the results. Since it may take some time to predict many sequences/many alleles the analysis can be run on existing predictions.
Usage largely involves setting up the config file and having your input files prepared. Running the command epitopepredict -c <yourfilename>.conf will create a new config file for you to work from if it doesn't exist. Just edit this with a text editor and then to execute:
epitopepredict -c <yourfilename>.conf -r
The advantage of configuration files is in avoiding long commands that have to be remembered or are prone to mistakes. Also the config files can be kept to recall what setting we used or to copy them for another set of files. The current options available in the file are shown below.
[base]
predictors = tepitope
mhc2_alleles = HLA-DRB1*01:01,HLA-DRB1*04:01
mhc1_alleles = HLA-A*01:01
mhc1_length = 11
mhc2_length = 15
n = 2
cutoff_method = default
cutoff = 4
sequence_file =
path = results
overwrite = no
verbose = no
names =
plots = no
genome_analysis = no
[iedbtools]
iedbmhc1_path =
iedbmhc2_path =
iedb_mhc1_method = IEDB_recommended
iedb_mhc2_method = IEDB_recommended
Settings explained:
| name | example value | meaning |
|---|---|---|
| predictors | tepitope | name of predictor can be: tepitope, iedbmhc1, iedbmhc2, netmhciipan, mhcflurry |
| mhc1_alleles | HLA-A*01:01,HLA-A*03:01 | list of MHC-I alleles or preset name |
| mhc2_alleles | HLA-DRB1*0101,HLA-DRB1*0103,HLA-DRB1*0401 | list of MHC-II alleles or preset name |
| mhc1_length | 11 | length of n-mers for MHC-I prediction |
| mhc2_length | 15 | length of n-mers for MHC-II prediction |
| n | 3 | minimum number of alleles for promiscuous binders |
| cutoff_method | default | cutoff method default |
| cutoff | 4 | percentile cutoff for counting promiscuous binders, i.e. top 4 percent |
| sequence_file | zaire-ebolavirus.gb | set of protein sequences in genbank or fasta format |
| path | results | folder to save results to, can be empty for current folder |
| overwrite | no | overwrite the previous results |
| names | Rv0011c,Rv0019c | protein/sequence/locus tag names to predict in your file, optional |
| verbose | no | displays more information while running |
| plots | yes | make plots of protein binders |
| genome_analysis | no | global analysis for all proteins |
| iedbmhc1_path | folder where the IEDB MHC-I tools are installed, not required unless used | |
| iedbmhc2_path | folder where the IEDB MHC-II tools are installed, not required unless used | |
| iedb_mhc1_method | IEDB_recommended | predictor to use within the IEDB MHC-I tools (see below) |
| iedb_mhc2_method | IEDB_recommended | predictor to use within the IEDB MHC-II tools (see below) |
For convenience there are some lists of common alleles that you can use without having to type allele names into the config file. These have been taken from various sources and are only a rough guide. Use epitopepredict -p to see the available presets.
The current selection is:
| name | description |
|---|---|
| mhc1_supertypes | 6 MHC-I supertypes |
| mhc2_supertypes | 7 MHC-II supertypes |
| us_caucasion_mhc1 | 30 most common US caucasion MHC-I |
| us_african_mhc1 | 30 most common US african MHC-I |
| human_common_mhc2 | 11 most prevalent HLA-DR alleles worldwide |
| broad_coverage_mhc1 | 26 alleles providing broad coverage |
| bovine_like_mhc2 | 8 HLA-DR alleles chosen to approximate bovine response |
ann
comblib_sidney2008
consensus
IEDB_recommended
netmhcpan
smm
smmpmbec
Using preset allele lists saves you the trouble of writing the alleles out. You can get the built-in presets by using -p at the command line. If you provide MHC-I alleles for a class II predictor like tepitope the program will give an error.
[base]
predictors = tepitope
presetalleles = common_human_mhc2
n = 2
cutoff = 5
sequence_file = zaire-ebolavirus.gb
path = results
names =
plots = yes
genome_analysis = no
In each results folder you will find csv files with the predictions for each sequence. This is the primary raw output. There is a separate folder for each prediction method. These folders can be re-used as input in the analysis section without re-running predictions.