CEDAR/Cancer Specific Rules

From curation_manual
Jump to: navigation, search

Cancer Specific Curation Rules

For cancer curation, follow all regular IEDB curations rules with the following rule additions. Note that these rules are still in process and are not yet comprehensive.

Cancer Antigens

Cancer antigens are defined as the epitope’s or ERO’s source Antigen when there is a positive B cell or T cell assay, and the host has cancer. If a protein is determined to be a cancer antigen, all references with epitopes or EROs from that protein are pulled into the CEDAR website

Occurrence of cancer

Do not use the in vivo process type of 'Occurrence of cancer' unless the epitope's source antigen is considered a cancer antigen. This means that the antigen is specifically expressed or overexpressed in cancer cells or tissue and was studied in that context. If the cancer type is a lymphoma (e.g. abnormal production of T cells or B cells / antibodies), characterizing the specificity of the T cells / antibodies produced by the tumor does not make the epitopes recognized 'cancer epitopes'. Note that cancer diseases may be relevant to the epitope and will not have Immunogens. Do capture these diseases whenever relevant. Always curate different diseases as separate assays.

Before curating the in vivo process type in a B cell or T cell assay as 'Occurrence of cancer', look at the epitope's source antigen and evaluate if the authors consider it to be a cancer antigen. If in doubt, ask for guidance.

  • For example, with Melanoma antigen recognized by T-cells 1 (MART1) epitopes studied in the context of melanoma, MART1 is considered a cancer antigen, as this protein is expressed specifically by the cancer cells.
  • For example, with myelin basic protein (MBP) epitopes studied in the context of  Waldenstrom’s macroglobulinemia, MBP is not considered a cancer antigen as myelin basic protein is NOT specifically expressed in the cancer cells.

Important Note: In cases with no relevant in vivo immunogen, such as cancer, when there is an additional in vitro restimulation with a relevant immunogen, be sure to capture the in vitro process as primary induction in vitro.


Neoepitope Definition

Multiple types of genetic changes occur in cancers leading to expression of new sequences. A mutation specifically found in a cancer patient is to be curated as a neoepitope, not an analog or mimotope.

subtypes of neo-epitope:

  • The epitope is an in-frame neo-epitope of:
  • The epitope is a frameshift neo-epitope of:
  • The epitope is a fusion neo-epitope of:
  • The epitope is an unspecified neo-epitope of:

In-frame neo-epitope

Definition: Mutation results in protein with small sections missing or substituted, but the rest of the protein is not changed. Will account for 90% of neoepitopes.


PMID: 30282837



PMID: 35122353



Required fields: only ERO fields

Neoepitope sequence (epitope object) 

Wildtype peptide sequence (epitope-related object)

Wildtype protein sequence (epitope-related object)

All other fields will be calculated.

Frameshift neo-epitope

Definition: Mutation disrupts the open reading frame. The protein sequence following the mutation is completely altered. will be a small percentage of total neoepitopes.

Example: PMID:24323902 Supp Table 1


Required fields (only if author provided):

Neoepitope sequence (epitope object) 

Wildtype peptide sequence (epitope-related object)

Wildtype protein sequence (epitope-related object)

The following fields are not required at this time, however, if the authors, provide this information, do enter it.

Details on how to complete these fields are TBD.

AA curated fields: mutation position, mutation aa, wild type aa

DNA curated fields: mutation chromosome, mutation position, mutation base, wild type base Gene Name calculated

Fusion neo-epitope

Definition: Product of gene fusion made by joining parts of two different genes.

Example: 35211402 Table 1


Required fields: only ERO

Neoepitope sequence (epitope object) 

Wildtype protein sequences (epitope-related object:2 proteins)

Gene Names calculated

Neoepitope Examples

  • When a peptide results from a tumor-specific mutation that disrupts the open reading frame of a protein through an insertion or a deletion (indel) of nucleotides into its coding region, curate this as a frameshift neoepitope (not SRC).  This includes peptides that overlap with the indel and those that are encoded downstream of the mutation.

Note: When in doubt (when it is not clear how the mutation arose) and if you are curating a cancer paper, curate it as a neoepitope.

  • If the epitope is derived from a tandem duplication of a natural peptide without any mutation, it is considered an unspecified neoepitope if the junction region is recognized.

Example: https://pubmed.ncbi.nlm.nih.gov/17119119/ A Neoepitope Generated by an FLT3 Internal Tandem Duplication (FLT3-ITD) Is Recognized by Leukemia-Reactive Autologous CD8+ T Cells Here the junction region is within the epitope, so these are curated as neoepitopes:

Neo junction ex.jpg

  • If the epitope is derived from fused gene products and spans the breakpoint, it is considered a fusion neoepitope.

Example: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC5831673/ Gene fusions are frequently found in prostate cancer and may result in the formation of unique chimeric amino acid sequences (CASQs) that span the breakpoint of two fused gene products. This study evaluated the potential for fusion-derived CASQs to be a source of tumor neoepitopes, and determined their relationship to patterns of immune signatures in prostate cancer patients.

PMID: 34120214 The MELOE-136–44 peptide (TLNDECWPA) is translated from a tumor-specific LncRNA. There is no ORF indel or splicing event involved, just tumor-specific transcription of a bicistronic LncRNA, from which the protein/peptide is translated. This is NOT curated as a neoepitope. It should be curated as a natural peptide, using external accession (UniProt, GenPept, PDB) or SRC, if no 100% blast hit exists.

Neoepitope Object Type

  • Neoepitopes with 100% blast match - Curate as peptide, no natural source
  • Neoepitopes without blast match - Curate as peptide, no natural source

Use the Epitope Related Object (ERO) fields to link the neoepitope to the wild type. Enter as much detail as the authors provide. If they give a peptide sequence, or it is easily deduced, always enter the wild type peptide as the ERO. If no peptide is available, the wild type protein may be used. In the worst case scenario, the ERO may be an organism.

ERO Issues

  • If authors claim a single specific mutation in the neoepitope, but all blast hits have multiple mutations or are 100% hits, then use the reference proteome peptide sequence as the ERO.

(PMID: 30880120)

  • When the ERO that is given by the author is a completely different peptide, with no homology to the neoepitope, use it anyway.

(PMID:30256815) https://www.ncbi.nlm.nih.gov/pmc/articles/PMC6157866/ Neo ERO ex3.jpg

  • If authors do not provide the wild type sequence and the neoepitope has only 100% blast hits in Reference Proteome = where is the mutation at?

(PMID: 30880120)

FYGKTILWF is neoepitope Reference proteome gives Q9H993 ARMT1_HUMAN Protein-glutamate O-methyltransferase OS=Homo sapiens OX=9606 GN=ARMT1 PE=1 SV=1 FYGKTIPWF = the wild type peptide ignore authors positions (most are wrong)

Neo ERO ex.jpg

  • If authors claim a single specific mutation in the neoepitope, but that position is not mutated in the reference proteome = which sequence to use as the epitope related object? (PMID: 30880120)

RYPRYLYKL, for which they say aa 1 is mutated, but they do not say what aa position 1 should be Ref proetome: A0A087X1B3_HUMAN Transmembrane protein 178B OS=Homo sapiens OX=9606 GN=TMEM178B RYPRYLYGL Use ref proteome seq, ignore authors positions

Neo ERO ex2.jpg

Analogs Mistaken for Neoepitopes

There are some commonly studied analogs that are often mistaken for neoepitopes. They occur in papers discussing neoepitopes where the authors do not clearly explain their origin and are often targets of epitope specific T cell lines where the nature of the epitope is not discussed.

The following peptides are ANALOGS


Important Note: When in doubt (when it is not clear how the mutation arose) and if you are curating a cancer paper, curate it as unspecified neoepitope. When in doubt if the epitope is a neoepitope, do not curate it as a neoepitope, curate as analog or use SRC, depending on if it was shown to be unnatural or natural.

Healthy Host

Curate assays where healthy controls are tested for recognition of neoepitopes and/or wild type sequences

Infectious Disease or Exposure Host

Do not curate cancer as the disease state for an infectious disease epitope unless that infection leads to cancer. For example, with an HTLV epitope, T cell lymphoma is a relevant in vivo process because HTLV infection is associated with T cell lymphoma. With an influenza epitope, melanoma is not a relevant in vivo process. With an HPV epitope, cervical cancer is a relevant in vivo process because HPV infection is associated with cervical cancer. With an influenza epitope, exposure to influenza virus or vaccine are the only relevant in vivo processes. The in vivo processes curated must always be relevant to the assay antigen.

When cancer is a result of an infectious process and the epitopes are from the infectious agent. IVP1 will then be ‘Occurrence of infectious disease’ with infectious disease filled and Immunogen = infectious agent, IVP2 will be ‘Occurrence of cancer’ with Immunogen = Blank.

  • Cancer status after infection will be recorded only if the patients all suffer from it or if there is a differential response to the epitope in diseased and non-diseased infected individuals.
  • If infected patients with and without the cancer are tested and the outcome is the same, they will be bulked under the infectious agent exposure alone.

Some viruses are commonly accepted as having a causal relationship with a specific cancer. This table shows some of these established relationships between a virus and a cancer:

Vir ca.jpg

Note: All child terms (virus strains and disease states) are always incuded. If you are curating a paper where an author explicitly states an association that is not in this table, curate according to the author's statement.

Animal Models of Cancer

  • Cancer in mice can be caused by injecting tumor cells that grow and mimic a natural cancer
  • Specific strains of mice spontaneously develop cancer by a certain age (TRAMP mice)

Both cases are to be curated as animal model of cancer, but if something is administered, the in vivo process type is administration and if the cancer occurs spontaneously, the in vivo process type is occurrence of cancer.

If you encounter a new animal model of cancer arises, use ‘animal model of cancer’ and request new child terms. Only those terms used to date are currently available.

Important Note: If the immunogen does not cause disease, do not fill in disease state in that process. For example, if a diseased human is injected with an epitope vaccine, that vaccine did not cause the disease, therefore do not enter a disease in the in vivo process describing the vaccination step. In this case, in vivo process 1 will be "Occurrence of cancer" with no immunogen and in vivo process 2 will be "Therapeutic Vaccination" with the vaccine as the immunogen.

Important Note: If tumor cells are injected at a site outside of the natural tumor site, the animal tumor is not a model of a specific human cancer. Instead, it is to be curated as 'animal model of cancer.' for example, if breast cancer cell lines are injected into the breast of an animal, that can be curated as 'experimental breast cancer', however, if the same cell lines are injected into the flank of the animal, that is to be curated as 'animal model of cancer.'

Important Note: 'Administration in vivo to cause disease' - Use of this process type depends on the assay type. Only in vivo assays utilize "to cause disease". Use in cases of experimental induction of a disease only in those assays that do not measure immune recognition directly but that, instead, record the existence of disease such as exacerbation and treatment assay types. For in vitro assay types, always use 'Administration in vivo' as the process type.

  • The assay has a hypothesis that the role of the immunogen is to cause a disease.


Vaccination is a subtype of Administration in vivo

It is to be used for vaccination with author described "vaccines", even when this was not administered in relation to the experiment being performed. Use Vaccination whenever the authors refer to the immunogen as a "vaccine" and it is not clear if it was administered prior to or after disease development.

Prophylactic vaccination is to be used when the vaccine is administered prior to disease development.

Therapeutic vaccination is to be used when the vaccine is administered after disease development.

Important Note: do not bulk these two vaccination types in cancer papers. In all other categories, they may be bulked under whichever protocol gave the best result.


Always capture the specific tissue that the tumor was derived from when describing effector cells, antigen presenting cells, immunogens, and antigens when they are derived from a tumor.

Generic “tumor” will not be added to the tissue types

If it is not known, Unknown/Other can be used


If metastatic cells are harvested to be used as effector cells or APC (regardless of cell type), the tissue type should be entered as either Metastatic Site or the tissue they were harvested from, depending on how relevant the cancer is to the immune recognition, not the tissue of the cancer’s origin. If the host's cancer is irrelevant, as with an influenza epitope, then the tissue they were harvested from is more important than the fact that the cells were metastatic. If the host's cancer is important, as with a neoepitope, then it is more important to capture the tissue as Metastatic Site.

For cancer relevant contexts:

if effector cells or APC are harvested from a resected metastatic tumor, the tissue field should be filled out as Metastatic Site.

if effector TILs are harvested from solid colorectal tumor metastases in the lung, the tissue type field is to be captured as Metastatic Site.

if APC are harvested from solid colorectal tumor metastases in the lung, the tissue type field is to be captured as Metastatic Site.


Always try to be as specific as possible when cancer cells are used as effector cells, antigen presenting cells, immunogens, and antigens. This means if the cells are CD8+ TILs, curate as CD8+ T cells, rather than as TILs because CD8+ is more specific than TIL.

Tumor Infiltrating Lymphocyte (TIL) is on the cell list

Make term requests when new cancer cells lines are encountered

Important Note: When different cell lines are used as antigen and if the cell type, tissue type, and importantly host and in vivo process type and disease are all the same, then bulk curate those different cell lines as a single assay antigen. Select the one cell line used most often, gives the best response, or is present in the drop down list.

Negative Monoclonal Responses

Always curate monoclonal negative data when it is deemed interesting, otherwise, it is not curatable. Lack of cross-reactivity to wild type qualifies as interesting. Curate wild type peptides as a negative assay antigen under the neoepitope if they are NOT recognized by a neoepitope monoclonal receptor. The wild type peptide is not curated as an epitope unless it is positive.

Keep in mind the overriding MHC binding rule, as stated here: http://curationwiki.iedb.org/wiki/index.php/Curation_Manual2.0#Add_MHC

Important Note: All experimental MHC binding data given in a reference will be entered in the database irrespective of whether the qualitative assessment is positive or negative. This may not be bulk curated under any circumstances.

Important Note: Do not curate an assay demonstrating that a mAb has no anti-tumor effects against cancer cell lines.


Treatments that that are unrelated to the epitope (e.g., vaccination with unrelated peptide or protein) are not curated as the immunogen. The immunogen must relate to the antigen being recognized. Treatments should be described in the Immunization Comments Field whenever they affect the recognition of the epitope and/or antigen.

in vivo Challenge Assays

Assay Antigen

In cancer challenge/treatment/exacerbation assays, where the cancer or tumor is naturally occurring, there is no clear assay antigen because the immune response occurs in vivo with the readout being either more or less disease. In these cases, the assay antigen will be curated as the host's equivalent of the epitope's or epitope related object's source antigen. This means if the epitope is from MART1 and the host is a mouse, the assay antigen is mouse MART1. If the epitope is from PMEL and the host is a human, the assay antigen is human PMEL.