Sieve-based relation extraction of gene regulatory networks from biological literature

BMC Bioinformatics

Table 3 Rule-based processing sieve input parameters.

Regular expression
Mention triplets Transcription	transcrib
Consecutive mentions
Transcription	.directs transcription.
Inhibition	.inactivate.
Inhibition	.inhibits.
Inhibition	.repressor to.
Inhibition ¹	.is negatively regulated by.
Activation¹	.is governed by.
Activation¹	.essential.activat.*
Activation	.to.activat.*
Activation	.turns on.
Requirement¹	.requires.
Requirement	.required.
Binding	.binds.to.*
Binding	-binding.*
List of consecutive mentions
Transcription	.under.control.of.
Activation¹	.is governed by.
Inhibition	.represses.
Inhibition	.to repress.
Sentences of consecutive mentions
Activation	.turns on.
Inhibition	.repressed.

Each of the four different rule-based extraction methods takes a target relation name and a regular expression as input. Some of them also require to specify whether the extraction should be made from active or passive sentences.
¹ The method is called with passive parameter set to true.

ISSN: 1471-2105