cancervariants / gene-harmony-analysis

gene-harmony-analysis

Background

Human gene symbols are regulated and follow guidelines established by HGNC. All genes are designated an authoritative symbol (also known as a primary gene symbol), a descriptive name, and an HGNC identification number. Although primary gene symbols are monitored to be unique, alias symbols are not.

Aliases are additional gene symbols and short descriptions that are used synonymously for the gene and/or any associated gene products. Gene symbols are curated from use in databases, experimental results, and literature. Primary gene symbols and aliases play a crucial role in referencing genes across publications, medical records, and data collections.

Our preliminary research uncovered collisions, or instances where a single gene symbol was used for multiple different genes. We have categorized them into two kinds:

a) Alias-primary collisions, which are gene symbols that are used as a primary gene symbol and an alias. The primary gene symbol KRAS is an alias in addition to a primary gene symbol.

b) Alias-alias collisions are gene symbols that represent an alias of multiple genes. The gene symbol VH is an alias for 35 genes in the NCBI database.

Out of the 43,164 genes in the HGNC database, 483 (1.12%) had alias-primary collisions and 2,084 (4.83%) had alias-alias collisions. The Ensembl database, which has 40,353 genes, was found to have alias-primary collisions in 218 (0.54%) of genes and alias-alias collisions in 3,680 (9.12%) of genes.

The NCBI database, which had the largest number of genes- 75,346, had 1,712 (2.27%) and 5,670 (7.53%) of genes with alias-primary and alias-alias collisions respectively, illustrating the prevalence of ambiguity that challenges the aggregation of genomic knowledge.

The total_alias_overlap Jupyter notebook shows the analysis to get these values (the notebook will be condensed and made more efficient)

Purpose

The difficulties associated with resolving ambiguity and ensuring accurate understanding of gene symbols restrict the rate of clinical decision-making and contribute to confusion in gene knowledge aggregation. The gene nomenclature system would be most effective if it is unambiguous with a tool to take existing knowledgebase entries as inputs to resolve.

This curated collection of collision data will be a foundation for disambiguating gene symbols.

How can you help?

Contributing information on collisions that you come across will help collect data on the collisions that would be most impactful to resolve as well as increasing the data available for developing resolution strategies for downstream tool development.

1. In the collision records folders, there are collision records that are completed (but can always be updated) and a blank sample record to use as a template.
2. The contributing documentation explains the different features that are included in a collision record.
3. To propose an update or make a new collision record create a personal fork to the repo. New collision records should be YAML files named after the collision in either the alias-alias or alias-primary collision records folders. Once created, the review process can be started with the creation of a pull request.

Contact Information

For any feedback, questions, or conversation, please make an issue.

gene-harmony-analysis's Issues

U4 alias-alias collision record add information field

Add a field for chromosomal gene location into record template. This will be useful for collision groups that have the same information in all other fields.

Make a notebook that adjusts all collision records if the structure of the record changes in the future

This Jupyter notebook should be able to update all of the collision records to a new structure when there are new developments.

Add more identifier number fields

Not all genes have an Ensembl ID. Need fields for HGNC and NCBI (Entrez) ID fields to be thorough

Automatic Fill Information Fields

I wonder if there is a way to help fill out collision records by providing a way to automatically fill out the ENSG id and chromosomal location fields.

Prune Collision Analysis Notebook

There is a lot of rough draft content that needs to be removed for easier reading and more efficient running

F379 alias-alias collision record missing collision acronym expansion

I was not able to find what F379 stood for when searching the literature

ALP1 alias-alias collision record missing collision source

The gene ASZ1 in the ALP1 collision group has an alias name of Ankyrin-Like Protein 1 but no linked publications that refer to it by that name.

Should there be a field for which source the collision is present in?

In the collision record, does it make sense, does it add value, to add a field that describes which source (NCBI, HGNC, or Ensembl) features that collision for that gene? For example, in the VH collision record, only NCBI Info uses VH as an alias and therefor the collision VH is only present in NCBI.

Add current alias-alias collision investigations into YAML file

The collision investigations are currently in a powerpoint slide deck. Need to transfer the info into a YAML

Provide guidelines for how to determine collision class and association

This framework should be easily changed as we come across new classes and associations. Some one should be able to use this document to decide what to put into the fields of collision_class and collision_association.