The applicability of this software

[MaryGoAround]

Hello

I have a set of vcf files (SNVs) from patients who have responded to chemotherapy versus a set of patients who have not responded.

I want to quantify the neoantigen load in the tumors ,can I still use your software?

In a paper I read: We first collected all peptides defined by a 17 amino-acid region centered on the amino acid which changes upon the mutation. We identified mutant nonamers with ≤500 nM binding affinity for patient-specific class I human lymphocyte antigen (HLA) alleles, constituting potential candidate neoantigens. Binding affinities were predicted using NetMHC-3.4 [82]. We then quantified the peptides that displayed high affinity binding in tumor, but low or no binding in the respective matched normal and obtained total counts

Is this almost the same which your software can do?

Thank you if you guid me if I am on the right track

[leeprichman]

Yes, as long as you have properly formatted VCFs and per sample HLA calls, antigen.garnish can be used for this kind of analysis!

[leeprichman]

Closing this for now, please reopen or open a new issue if you have further questions!

[MaryGoAround]

Sorry, does your software need HLA type of a given patient (.VCF) ? I mean your package don't use HLAs?

I only have access to .vcf (SNVs and INDELs separately) coming from whole genome sequencing

Is this adequate for your software to predict Neoantigen load?

I got worry because I heard HLA should be patient specific

Some tools like Optitype package predict HLA types but does need fastq or bam format (if I am not wrong) which I don't have access to Please give an intuition

Thank you

[leeprichman]

antigen.garnish does not predict HLA alleles. You must know your alleles of interest for each VCF/patient in order to perform neoantigen analysis.

VCF files only contain lists of variants. They do not contain enough information to determine HLA haplotypes. You will need to use the sequencing data to determine HLA types per patient prior to predicting neoantigen burden with any pipeline.

[MaryGoAround]

Sorry in which part of your code, your software demands alleles of interest for each VCF/patient?

Can not I use all of human alleles instead. (HLA-A, HLa-B,HLA-C,and class II) ?

Because in this part of your script says

# MHC may also be set to "all_human" or "all_mouse" to use all supported alleles

dt[, MHC := c("HLA-A*01:47 HLA-A*02:01 HLA-DRB1*14:67")]

Thank you for any help

[leeprichman]

You could run all possible MHC alleles but it would be computationally quite intensive and wouldn’t really give you neoantigen burden. Neoantigen burden is determined by the predicted peptides derived from tumor mutations that have affinity for the subjects specific MHC alleles.

If you are just interested in any possible mhc allele being capable of presenting one of your mutant peptides, then this type of analysis could be useful.

[MaryGoAround]

Actually, from this paper https://www.ncbi.nlm.nih.gov/pmc/articles/PMC5957269/ I noticed they are using this approach just using these HLAs

HLA-A01:01
HLA-A01:04
HLA-A02:01
HLA-A02:06
HLA-A02:89
HLA-A03:01
HLA-A03:21
HLA-A03:26
HLA-A11:01
HLA-A23:01
HLA-A24:02
HLA-A24:03
HLA-A25:01
HLA-A26:01
HLA-A29:01
HLA-A29:02
HLA-A30:01
HLA-A30:02
HLA-A31:01
HLA-A32:01
HLA-A33:01
HLA-A68:01
HLA-B07:02
HLA-B07:05
HLA-B08:01
HLA-B13:02
HLA-B14:02
HLA-B18:01
HLA-B27:05
HLA-B35:01
HLA-B35:02
HLA-B35:12
HLA-B35:40
HLA-B37:01
HLA-B38:01
HLA-B40:02
HLA-B400:102
HLA-B41:01
HLA-B44:02
HLA-B44:03
HLA-B450:101
HLA-B48:01
HLA-B49:01
HLA-B51:01
HLA-B51:08
HLA-B55:01
HLA-B56:01
HLA-B57:01
HLA-C01:02
HLA-C02:02
HLA-C03:03
HLA-C03:04
HLA-C04:01
HLA-C05:01
HLA-C06:02
HLA-C07:01
HLA-C07:02
HLA-C07:04
HLA-C08:02
HLA-C08:03
HLA-C12:03
HLA-C15:02
HLA-C15:04
HLA-C15:05
HLA-C16:01
HLA-C16:02
HLA-C17:01

I emailed the author, she says that

I don’t remember the details anymore, but the HLA types are definitely not predicted from the cohort. I can see if I can find this information, but netMHC documentation should explain how to get this information. You can probably use the same file, but do a bit of a search though in case there have been some updates on HLA types in human in the past 4 years.

[leeprichman]

If you have vcf files, you can run antigen.garnish with whichever MHC alleles you are interested in.

[beginner984]

Sorry, please can you suggest/refer me to a common HLA types in Human for predicting Neoantigen load as I don't have access to HLA typing specific to my patients

Can I simply use your all_alleles.txt from https://github.com/immune-health/antigen.garnish/blob/master/inst/extdata/all_alleles.txt if that is a combination of mhcflurry_alleles.txt , netMHCII_alleles.txt , netMHCIIpan_alleles.txt , netMHC_alleles.txt and netMHCpan_alleles.txt

I checked but there was not any thing common between `all_alleles.txt and the other allele files

Am I right?

Thank you so much for any help

[leeprichman]

I don’t have a suggestion for this. I’m sorry. I don’t know if any resource with a consensus for common alleles

[beginner984]

Sorry is your all_alleles.txt a combination of mhcflurry_alleles.txt , netMHCII_alleles.txt , netMHCIIpan_alleles.txt , netMHC_alleles.txt and netMHCpan_alleles.txt ?

[leeprichman]

Yes.