Ligand.MHC Atlas: Guide

✦ 1. Database content

1.1 Overview of Ligand.MHC Atlas

The identification of (neo)antigens is of paramount importance for the development of effective cancer immunotherapies. Neoantigens, derived from tumor-specific mutations, can be recognized by the immune system and serve as critical targets for personalized cancer treatments. Accurate identification and characterization of these neoantigens can significantly enhance the efficacy of immunotherapies, such as checkpoint inhibitors, adoptive T cell therapies, and personalized cancer vaccines, by improving their ability to target cancer cells while sparing healthy tissues. MS-based immunopeptidomics, the large-scale study of peptides presented by MHC alleles, has become an essential field for understanding tumor immunology. For instance, adoptive T cell therapies, such as T cell receptor (TCR) transgenic T cells and chimeric antigen receptor (CAR) T cells, rely on the precise identification of tumor-specific peptides to enhance their targeting capabilities and efficacy. Moreover, understanding the immunopeptidome can aid in the prediction of patient responses to checkpoint inhibitors, thereby optimizing treatment strategies. More than a dozen mutation-derived neoantigens have been reported from immunopeptidomic analysis of cancer cell lines and tumor samples.

Ligand.MHC Atlas:

◆ We systematically processed HLA class I and II immunopeptidomes from 112 studies, and yielded 24,380,595 and 7,300,227 peptide-spectrum matches (PSMs) from 305.8 million MS/MS spectra.

◆ We constructed the Ligand.MHC Atlas, which contained 710,905 HLA-I and 209,125 HLA-II bound immune peptides covering 292 HLA alleles, from 26 major human cancers (54 subtypes).

◆ We provided a comprehensive landscape of PTM antigens, including 414,176 tumor-associated PTM antigens.

Figure 1. Content and statistics of the Ligand.MHC Atlas.

1.2 Data Processing Workflow

Figure 2. The workflow and analysis pipeline of Ligand.MHC Atlas.

1.2.1 Immunopeptidomic datasets collection

The Ligand.MHC Atlas emphasizes the comprehensiveness in cancer immunopeptidomic datasets collection. It includes an extensive collection of more than 7,000 mass spectrometry (MS) raw files sourced from 114 published immunopeptidomic datasets.Data collection is sourced from multiple repositories, including PRIDE [1], MassIVE [2], PeptideAtlas [3], JPOST [4], and iProX[5]. Once the datasets are identified, they can be downloaded either from FTP sites or directly from web pages. We have downloaded over 10 terabytes of raw data.

1.2.2 Peptides identification

After collecting the RAW files, we used the reviewed dataset of 20,435 human proteins and 288 contaminants (such as keratins) downloaded from UNIport as the search database for peptide identification. We employed the search software pFind 3.2.1[6], which has been validated to identify the most peptides[7]. The search strategies for low-precision and high-precision data differed to ensure data accuracy. High-precision data was processed using an open search mode, while low-precision data utilized a closed search mode. We applied a stringent criterion with an FDR (False Discovery Rate) of less than 1% to evaluate the results at the PSM (Peptide-Spectrum Match), peptide, and protein levels. Overall, we identified 253.7 million MS/MS spectra, yielding a total of 2,367,690 unique ligands.

1.2.3 Sample information collection

The Ligand-MHC Atlas emphasizes the annotation of sample information, and in fact, one of our highlights is that users can retrieve peptide information for individual samples, which provides a more precise reference value. By reading the original literature of the production dataset, we annotate a sample from four aspects, which are Sample, Disease Type, Sample Source, and Sample Type.

Regarding the samples, an important aspect is the collection of HLA Class I and Class II allele information from patient samples, which is utilized for subsequent predictions by machine learning tools.

1.2.4 Peptide deconvolution, binding prediction

For HLA-I: MixMHCpred 2.0.2 and MHCflurry 2.0 followed the binding classification criteria of NetMHCpan 4.1, Only ligands between 8 and 14 amino acids were considered. The %Rank value was primarily used to assess binding scores.

Since different models have their own strengths and limitations, setting the threshold at ≥2 ensures that at least two tools predict strong binding, improving binding event accuracy. The integration of multiple prediction tools enhances the reliability and robustness of our deconvolution process by reducing potential biases from individual models. Additionally, we also utilized NetMHCpan 4.1 and NetMHCIIpan 4.0[3] to generate binding affinities, which are available on our web interface as an additional reference for users.

✦ 2. Usage

2.1 Quick start in the Ligand-MHC Atlas

In the Ligand.MHC Atlas database, interactive interfaces were constructed in a user-friendly manner.

1) Main functions of the database are provided in menu bar form.

2) Input any field we provide, for quick search.

3) Two examples of peptide-HLA-allele pairings to quickly familiarize users with the content of the Ligand-MHC Atlas.

2.2 Five types of search modes in the Ligand-MHC Atlas

Five search options were designed and developed to obtain the associations of immunopeptidome and HLA alleles, including 1) Simple search, 2) Advanced search, 3) Immunopeptide search, 4) HLA search, and 5) Protein search.

1). Simple search:

Example: Please click on the "Example" button to search "FLDETLRSLA" in the "Immunopeptide" field. After click on the "Submit" button, the related immunopeptides will be shown.

2). Advanced search:

Example: Please click on the "Example" button to search the interactions of "FLDETLRSLA" and "HLA-A*02:01". By clicking on the "Submit" button, the related information will be shown.

3). Immunopeptide search:

Example: Please click on the "Example" button to load an example, which seach "FLDETLRSLA". The related immunopeptides will be shown after clicking on the "Submit" button.

4). HLA search:

Example: Please click on the "Example" button to load "HLA-A*02:01". By clicking on the "Submit" button, the related HLA alleles will be shown.

5). Protein search:

Example: Please click on the "Example" button to seach "GFRA2". By clicking the "Submit" button, the related information will be shown.

2.3 Browse in Ligand-MHC Atlas

Four browse options were provided to acquire the core dataset in the Ligand.MHC Atlas, contains 1) Cancer browse, 2) HLA browse, 3) specific Ligand browse, and 4) PTM browse.

1). Browse by Cancer:

You can select one type of cancers to browse all MHC-bound peptides of corresponding cancer type.

1) Click here to view cancer specific peptides.

2) Click here to view peptides specific to a certain type of tumor and its corresponding HLA-allele.

2). Browse by HLA:

You can select one of these HLA molecules to browse all bound peptides of corresponding HLA.

1) Click here to view HLA-alleles specific peptides.

2) Click here to view peptides for a specific HLA-allele and a particular sample.

3). Browse by Ligand:

You can directly click on the peptide ID to navigate to the annotation page in this option.

4). Browse by PTM:

You can choose a PTM that you are interested in and click to browse the relevant content and search for your target item.

✦ 3. Basic information of a immunopeptide

To provide the basic information of a peptide segment, the following details should be included: the peptide's identifier, the number of HLA alleles associated with the peptide, the source protein from which the peptide is derived, the BY ion spectrum.

1) Peptide Identifier: This is the unique identifier for the peptide within the Ligand-MHC Atlas.

2) Number of HLA Alleles: This refers to the quantity of HLA alleles that the peptide segment is capable of binding to.

3) Protein of Origin for the Peptide: This includes information such as the UniProt number, gene name, protein name, and the specific location of the peptide within the protein sequence from which the peptide is derived.

4) The network diagram with the peptide at the center, connected by arrows to its corresponding HLA alleles, which further point to related cancer types.

5) BY Ion Plot: Utilized for the identification and sequence determination of the peptide segment, it offers the corresponding image downloads and PSM-scores.

6) The dataset and sample from which the peptide originates.

7) Predicted information regarding the binding affinity between the peptide and HLA molecules.

✦ 4. References

[1] PEREZ-RIVEROL Y, BAI J, BANDLA C, et al. The PRIDE database resources in 2022: a hub for mass spectrometry-based proteomics evidences [J]. Nucleic Acids Res, 2022, 50(D1): D543-d52.

[2] CHOI M, CARVER J, CHIVA C, et al. MassIVE.quant: a community resource of quantitative mass spectrometry-based proteomics datasets [J]. Nat Methods, 2020, 17(10): 981-4.

[3] OMENN G S, LANE L, OVERALL C M, et al. The 2023 Report on the Proteome from the HUPO Human Proteome Project [J]. J Proteome Res, 2024, 23(2): 532-49.

[4] WATANABE Y, YOSHIZAWA A C, ISHIHAMA Y, et al. The jPOST Repository as a Public Data Repository for Shotgun Proteomics [J]. Methods Mol Biol, 2021, 2259: 309-22.

[5] MA J, CHEN T, WU S, et al. iProX: an integrated proteome resource [J]. Nucleic Acids Res, 2019, 47(D1): D1211-d7.

[6] CHI H, LIU C, YANG H, et al. Comprehensive identification of peptides in tandem mass spectra using an efficient open search engine [J]. Nat Biotechnol, 2018.

[7] YI X, LIAO Y, WEN B, et al. caAtlas: An immunopeptidome atlas of human cancer [J]. iScience, 2021, 24(10): 103107.

✦ 5. Contact

✦ Dr. Shaofeng Lin: linshaofeng-at-fjmu.edu.cn.

✦ Dr. Haodong Xu: Haodong.Xu-at-uth.tmc.edu.

Ligand.MHC Atlas

a comprehensive atlas of MHC immunopeptidome in human cancer

Version 1.0

1. Workfolw »

2. Usage »

3. Information »

4. References »

5. Contact »