PSI-MITAB interactions

From ReactomeWiki

Jump to: navigation, search

What is PSI-MITAB?

The HUPO Proteomics Standards Initiative (PSI) defines community standards for data representation in proteomics to facilitate data comparison, exchange and verification.

One of its workgroups deals with Molecular Interactions (PSI-MI), with MITAB being their tab delimited data exchange format.

Interactions in Reactome

The Reactome database doesn't primarily focus on molecular interactions, but such data can be derived from the molecular reactions and complexes annotated in Reactome.

Briefly, two molecules are reported as interactors in Reactome, if they

a) are involved in the same Complex/Polymer or bona fide Reaction - the interaction type for these is "physical association"

b) are involved in neighbouring Reactions or associated via (positive or negative) Regulation - the interaction type for these is "association"


For more details see also the Reactome-specific PSI-MITAB definition below, especially columns 12, and 17 to 19.

The Reactome-specific PSI-MITAB format definition

The current MITAB formatted interaction file can be downloaded from the Reactome download page.


The column contents is as follows (Reactome specific notes indented):

1. Unique identifier for interactor A, represented as databaseName:ac, where databaseName is the name of the corresponding database as defined in the PSI-MI controlled vocabulary, and ac is the unique primary identifier of the molecule in the database. Identifiers from multiple databases can be separated by "|". It is recommended that proteins be identified by stable identifiers such as their UniProtKB or RefSeq accession number.

2. Unique identifier for interactor B.

  • Reactome uses UniProt accessions for all human annotation.

3. Alternative identifier for interactor A, for example the official gene symbol as defined by a recognised nomenclature committee. Representation as databaseName:identifier. Multiple identifiers separated by "|".

4. Alternative identifier for interactor B.

  • This holds the gene names for the UniProt accessions.

5. Aliases for A, separated by "|". Representation as databaseName:identifier. Multiple identifiers separated by "|".

6. Aliases for B.

  • None for now.

7. Interaction detection methods, taken from the corresponding PSI-MI controlled vocabulary, and represented as darabaseName:identifier(methodName), separated by "|".

  • "inferrred by curator" for all Reactome interactions.

8. First author surname(s) of the publication(s) in which this interaction has been shown, optionally followed by additional indicators, e.g. "Doe-2005-a". Separated by "|".

9. Identifier of the publication in which this interaction has been shown. Database name taken from the PSI-MI controlled vocabulary, represented as databaseName:identifier. Multiple identifiers separated by "|".

  • Reference describing the Reactome project for all Reactome interactions.

10. NCBI Taxonomy identifier for interactor A. Database name for NCBI taxid taken from the PSI-MI controlled vocabulary, represented as databaseName:identifier. Multiple identifiers separated by "|".

Note: In this column, the databaseName:identifier(speciesName) notation is only there for consistency. Currently no taxonomy identifiers other than NCBI taxid are anticipated, apart from the use of -1 to indicate "in vitro" and -2 to indicate "chemical synthesis".

11. NCBI Taxonomy identifier for interactor B.

12. Interaction types, taken from the corresponding PSI-MI controlled vocabulary, and represented as dataBaseName:identifier(interactionType), separated by "|".

  • "physical association" for all interactors occurring in the same complex, polymer or in the same reaction in Reactome (see columns 17 and 18)
  • "association" for interactors involved in neighbouring reactions or associated via (positive or negative) regulation in Reactome (see columns 17 and 19).

Note that 'associations' are only recorded for identifier pairs that are not listed under 'physical association'.

13. Source databases and identifiers, taken from the corresponding PSI-MI controlled vocabulary, and represented as databaseName:identifier(sourceName). Multiple source databases can be separated by "|".

  • "reactome" for all Reactome interactions.

14. Interaction identifier(s) in the corresponding source database, represented by databaseName:identifier

  • Reactome doesn't have identifiers for interactions, so this is empty.

15. Confidence score. Denoted as scoreType:value. There are many different types of confidence score, but so far no controlled vocabulary. Thus the only current recommendation is to use score types consistently within one source. Multiple scores separated by "|".

  • Not applicable, so this is empty.

All columns up to here are mandatory.

Additional columns for the Reactome PSI-MITAB format:

16. Id of the pathway(s) that include(s) the interaction. Stable id when available, otherwise internal db_id. Multiple ids separated by "|". Only the pathway(s) that directly contain(s) the reaction(s) where the interaction can be identified are recorded here, no upstream pathways.

17. Id of the reaction(s) that include(s) the interaction. Stable id when available, otherwise internal db_id. Multiple ids separated by "|".

Four "types" of reaction contexts are recorded here:
a) Reactions that contain complexes or polymers that include the interaction. These reactions can be BlackBoxEvents.
b) Reactions that involve entity pairs that don't reside in the same complex or polymer. These need to be bona fide Reactions, i.e. BlackBoxEvents are excluded here.
Reactions of type a or b are listed under 'physical association'.
c) Pairs of neighbouring reactions. Here the interactors are involved in neighbouring reactions, and the respective reactions are listed in the same order as the interactors.
d) Reactions implicated in regulation instances, either directly as regulators or regulated entities, or via a catalyst activity. The corresponding regulation id is listed in column 19.
Reactions of type c or d are listed under 'association'.

Note: Entities that are members of the same EntitySet in a reaction (and don't re-occur in a different context in the reaction) are not recorded as interactors with each other.

18. Id of the complex(es) or polymer(s) that include(s) the interaction. Stable id when available, otherwise internal db_id. Multiple ids separated by "|".

Interactors can reside directly within a given complex, or within one or more subcomplexes of a given complex. Interaction of an entity with itself is only recorded when this occurs directly within a given complex or polymer, or within *one* subcomplex (or "subpolymer") of a given complex.
Note: Entities that are members of the same EntitySet within a complex (and don't re-occur within the complex) are not recorded as interactors with each other.

19. Id of the regulation instance(s) that include(s) the interaction. Stable id when available, otherwise internal db_id. Multiple ids separated by "|". Reactions associated with the regulation(s) are listed in the reaction column - these are either reactions directly implicated as regulators or regulated entities, or reactions implicated via catalyst activities that are regulators or regulated entities.

Where values are missing, an empty cell is marked by "-".