MHC class I antigen processing consists of multiple steps that result in the presentation of MHC bound peptides that can be recognized as T cell epitopes. Many of the pathway steps can be predicted using computational methods, but one is often neglected: mRNA expression of the epitope source proteins. We improve epitope prediction by taking into account both peptide-MHC binding affinities and expression levels of the peptide’s source protein. Specifically, we utilized biophysical principles and existing MHC binding prediction tools in concert with experimental RNA expression values to derive a function that estimates the likelihood of a peptide being presented on a given MHC class I molecule. Our combined model of Antigen eXpression based Epitope Likelihood-Function (AXEL-F) outperformed predictions based only on binding or based only on antigen expression for discriminating eluted ligands from random background peptides as well as in predicting neoantigens that are recognized by T cells.

Axel-F can accept either CSV file or FASTA formatted file. Note that when providing CSV file as an input for Axel-F, it must be in a valid CSV format. A valid CSV file will contain a header and have no missing data or empty cell in each row. As for FASTA formatted input, it should be a valid format that agrees with the NIH standards.


There are multiple ways of what to include inside the CSV file. The simplest form of data you can provide to Axel-F is a CSV file containing peptide sequences only. Depending on available data you have, you may include allele names, tpm values, and/or gene names at your choice.


Please note that these are CSV formatted file, thus TSV format would not work. If you would like to simply create CSV formatted file on a text editor or directly on the form, please have each column separated by comma.


Because Axel-F requires allele and TPM value, you must specify them on the form if not already included in the CSV file. Let's take Example 1, which includes peptide sequences only. Axel-F would require you to specify allele and TPM value on the form. If Example 2 was provided, you don't need to specify allele on the form, but you still do need to provide TPM value. If Example 3 was provided, you don't have to specify anything since required information (Peptide sequence, allele names, TPM values) were all provided inside the CSV file.

Using Example 1, you can either manually write out the sequences in the textarea with the header "peptide" included, or drag and drop a CSV file containing peptide sequences onto the textarea. If you want to include any other data, make sure you include proper header on top, such as "allele", "tpm", or "gene name". Axel-F requires at least one allele to be specified in order to run the prediction. Unless you provided "allele" column in the CSV file, please select species of interest and appropriate allele from the dropdown menu on the form. If you have a custom TPM value in mind, please enter the value in the textfield or include it in the CSV file. Otherwise, you may specify TPM value in the form. Also note that TPM value must be an integer or decimal value. If you would like to get TPM value derived from TCGA data instead, click on "TCGA" button for "Select TPM source" and specify both cancer type and gene name. If the CSV file already contains list of gene names, you only need to specify cancer type on the form.

As you type in gene name, it will also autosuggest or recommend available genes.

FASTA formatted input is a lot more direct compared to CSV file as there really only one type of format that Axel-F can accept. A valid FASTA format is single-line description starting with ">" character, followed by lines of sequence data.
** Note : Currently Axel-F can accept only one FASTA sequence at a time.

If you fill the textbox with FASTA formatted data, AXEL-F form will detect that it's a FASTA format and will change available options. Compared to CSV format options, you can see from the following image that the form now has a length option added to it.

You can either manually write out the sequences in the textarea in proper FASTA format (starting with signle-line description starting with ">" character) or drag and drop an existing FASTA file onto the textarea. Unlike with CSV examples, predictions are limited to one specific binding length. Selecting one length will be applied to all peptides from the FASTA sequence. Axel-F requires at least one allele to be specified in order to run the prediction. Please select species of interest and appropriate allele from the dropdown menu on the form. If you have a custom TPM value in mind, please enter number in the textbox. Also note that TPM value must be an integer or decimal value. If you would like to get TPM value derived from TCGA data instead, click on "TCGA" button for "Select TPM source" and specify both cancer type and gene name.

Continuing with Example 1 from CSV input, after submitting the form, you should get the following results.

Rank EL values are directly from the neural networks that NetMHCpan uses. Because these numbers are abstract and cannot be directly used in biological context of Axelf, Rank EL values are translated to IC50 values by comparing the percentile ranks of the two metrics in Trolle dataset, then using interpolation function to map each percentile ranks to corresponding IC50 value. The resulting values are stored in Rank_mapped_to_IC50 column. The same interpolation function based in the Trolle dataset was used to calculate AXEL-F scores.

AXEL-F scores estimates the likelihood of a peptide being presented on HLA and being an epitope.