Skip to main content
Figure 1 | Journal of Biomedical Semantics

Figure 1

From: Sequential pattern mining for discovering gene interactions and their contextual information from biomedical texts

Figure 1

General framework to extract gene interactions. Figure 1 presents the overall process to detect and characterize gene interactions. There are two steps. The first step is the extraction of patterns. Sequential patterns are mined from a learning corpus that contains sentences representing gene interactions. In order to reduce the number of extracted patterns, constraints and recursive mining are applied. At the end, few sequential patterns corresponding to candidate linguistic interaction or characterization rules remain. A key point is that the sequence of words expressing the interaction in a pattern is automatically discovered. As an example, the sequence of words <AGENE, "association" > in the pattern “AGENE association with AGENE” (see Table 5) where a small gap may appear between AGENE and "association" are discovered by our method. This pattern conveys an interaction between two genes (denoted AGENE) in association. The sequential patterns are then analyzed and validated by experts. The ones that are not relevant for interaction detection or characterization are removed. The second step is the application of those validated patterns as linguistic rules to discover and characterize interactions in new corpora.

Back to article page