Send us your feedback. Sign up for Pathway Commons announcements.

Exporting to the Simple Interaction Format (SIF)

Introduction:

By default, all interaction networks and pathways in Pathway Commons are stored in the BioPAX exchange format. However, a number of web service commands, including get_neighbors and get_record_by_cpath_id, support the export of interaction networks to the Simple Interaction Format (SIF). SIF was originally created for use with Cytoscape, the open source bioinformatics software platform for visualizing molecular interaction networks. SIF is simple to parse, and easy to load into Cytoscape and other third-party applications.

SIF Format:

The SIF export from Pathway Commons is formatted as:

physical_entity_id <relationship type> physical_entity_id

where physical_entity_id is a valid primary ID in Pathway Commons, and <relationship type> is one of the interaction inference rules specified below.

Inference Rules:

It is not possible to fully translate a BioPAX formatted network (which is capable of storing rich biological semantics, including multi-participant relationships) into a SIF binary network (which is only capable of storing pairwise interactions between physical entities). We therefore infer pairwise SIF relationships from more complex BioPAX relationships using a set of rules. In general, this translation is lossy. Nonetheless, this SIF network remains useful for certain types of bioinformatic applications that require pairwise interaction input.

For those commands that support SIF export, users can specify an optional binary_interaction_rule argument. This is a comma separated list of inference rules (see table below) that are applied when binary interactions are requested. If not specified, all binary interaction rules will be applied.

binary_interaction_rule	Original BioPAX Interaction	Inferred Set of Pairwise Interaction
COMPONENT_IN_SAME: This rule infers an interaction if the first entity is a component of the second entity, which is a complex. This interaction is transient in the sense that A component_of B and B component_of C implies A component_of C. This interaction is undirected.

COMPONENT_OF: This rule infers an interaction if two entities belong to at least one molecular complex. This does not necessarily mean they interact directly. In a complex with n molecules, this rule will create a clique composed of n(n-1)/2 interactions. This interaction is directed.

CONTROLS_METABOLIC_CHANGE: This rule infers an interaction if the first entity catalyses a reaction that either consumes or produces the second entity. More specifically, this rule will find an interaction between two entities A and B if and only if A controls a conversion which B participates and appears only on the left or right side of the conversion but not both. This interaction is directed.

CONTROLS_STATE_CHANGE: This rule infers an interaction if the first entity catalyses a reaction that either consumes or produces the second entity. More specifically, this rule will find an interaction between two entities A and B if and only if A controls a conversion which B participates and appears both on the left or right side of the conversion. This interaction is directed.

CO_CONTROL_DEPENDENT_ANTI: This rule infers an interactionif the first and second entities have control over the same process, their control is dependent, i.e. one of them have effect over control of the other one, and their effect is in different directions (one of them activates, the other inhibits). This interaction is undirected

CO_CONTROL_DEPENDENT_SIMILAR: This rule infers an interactionif the first and second entities have control over the same process, their control is dependent, i.e. one of them have effect over control of the other one, and their effect is in the same direction (both activates or both inhibits). This interaction is undirected

CO_CONTROL_INDEPENDENT_ANTI: This rule infers an interaction if the first and second entities have control over the same process, their control is independent, i.e. they act without affecting each other's activity, and their effect is in different directions (one of them activates, the other inhibits). This interaction is undirected

CO_CONTROL_INDEPENDENT_SIMILAR: This rule infers an interaction if the first and second entities have control over the same process, their control is independent, i.e. they act without affecting each other's activity, and their effect is in the same direction (both activates or both inhibits). This interaction is undirected

PARTICIPATES_CONVERSION: This rule infers an interaction if both A and B participates in a conversion as substrates or products. Controllers are not included. This interaction is undirected.

PARTICIPATES_INTERACTION: This rule infers an interaction if both A and B participates in an interaction as participants. Controllers are not included. This interaction is undirected.

SEQUENTIAL_CATALYSIS: This rule infers an interaction if A and B catalyzes two conversions that are connected via a common molecule, and where potentially that common substrate is produced by the former and consumed by the latter. This interaction is directed

Legend

Retrieving Physical Entity Details:

In the SIF export defined above, physical entities are identified with physical_entity_id(s), and not with standard gene names or external identifiers. To retrieve the identity of each physical entity, you must make a Pathway Commons and use the get_record_by_cpath_id. In this case, you specify one or more internal IDs, and set output=biopax. This enables you to retrieve the full physical entity information, in BioPAX format.