jmillanacosta / omicsfixid Goto Github PK

Biological entities such as genes, proteins, complexes, and metabolites often have diverse identifiers across various databases, which pose a challenge for data integration. To solve this problem, identifier mapping is required to convert identifiers from one database to corresponding entities from other databases.

After mapping identifiers from one database to another, integrating data can remain a challenge due to the presence of retired, deleted, split, and/or merged identifiers currently present in databases and datasets alike.

Many biological databases split/merge or withdraw identifiers. Below you can see some examples from HGNC:

Some molecular entries were withdrawn/deleted from a database and they won't exist anymore. HGNC:440 is an example of withdrawn id from HGNC.

When an id is split or merged in a database, a new id(s) will be used for that specific entity. The new id(s) is called the primary id(s), while the split/merged id is the secondary id, which will not be used anymore.

The split ids may introduce one-to-multiple mapping issues which should be further evaluated.

In some databases, multiple ids refer to the same entity. We define these ids as duplicate ids. Below you see an example from HMDB:

In this case, the id, currently used by the databases to refer to the entity, is the primary id.

These outdated identifiers are called “secondary” while the identifiers currently supported by databases are referred to as “primary”. The presence of secondary identifiers in a used dataset or database can lead to information loss and hinders effective data integration. While some tools exist to convert secondary identifiers to current ones, these tools only support one type of data (either genes/proteins or metabolites):

• https://www.genenames.org/tools/multi-symbol-checker/
• https://www.metaboanalyst.ca/MetaboAnalyst/upload/ConvertView.xhtml

These tools currently do not have an API or other form of programmatic access, leading to issues in big OMICS data analysis.

To address the challenges of integrating data from different biological sources that contain secondary identifiers, we developed a user-friendly Shiny app called omicsFixID, which provides two key functions:

1. FixID: provides statistics on the percentage of secondary identifiers in the dataset and converts outdated secondary identifiers to current primary identifiers, if available. The FixID functionality currently covers secondary identifiers from HGNC , HMDB , ChEBI and Wikidata which can be converted to the corresponding primary identifier from the initial database. After this step, the IDMapper can be used to convert the primary-ID-enhanced dataset to any other database currently supported by BridgeDb:

• The full overview of supported databases is available on the BridgeDb website (bridgedb.org/pages/system-codes).

2. IDMapper: uses BridgeDb's REST-API to convert identifiers.

The metadata for the latest update of the mapping files is also available to users for queries within the app.

Future development entails updating the Secondary-to-Primary identifier mapping files regularly via GitHub Actions to ensure accuracy.