0%

​

Share this page with:

Predicting protein signature relationships

It is possible to automatically predict relationships between signatures by calculating the Jaccard index (a statistic used for comparing the similarity of sample sets, in this case protein matches) for each pair of signatures. This allows us to produce a predicted tree of relationships. An example of this semi-automatically generated tree is shown below (Figure 3) for a subset of the SFLD superfamily ‘Enolase’. While this system functions as a useful preliminary guide for InterPro curators, manual curation of most relationships between signatures is still required.

Figure 3 Semi-automatically generated hierarchical tree of relationships based on comparisons of proteins matched by each signature. Dashed lines indicate a lower confidence of the predicted relationship. The domain architectures shown represent the dominant architecture amongst protein matches for that signature. Functional information (including EC groupings) about matching proteins is also shown on the right of the image. An * following a signature name indicates the presence of further equivalent signatures at that node.