{% extends "base.html" %} {% import "bootstrap/wtf.html" as wtf %} {% import "bootstrap/fixes.html" as fixes %} {% import "bootstrap/utils.html" as util %} {% block title %}About{% endblock %} {% block scripts %} {{ super() }} {% endblock %} {% block content %}
{{ util.flashed_messages(dismissible=True, container=False) }}

Curation Protocol

As an application scenario, ComPath was used to generate mapping files across pathway databases in order to establish the multiple relationships between equivalent or related pathways in three of the major pathway databases (KEGG, Reactome, and WikiPathways). The protocol established for the mapping is described below.

Mapping Files

A total of 3 mapping files were curated, one for each pairwise comparison:

  • KEGG vs Reactome
  • KEGG vs WikiPathways
  • Reactome vs WikiPathways

Description

For each pairwise pathway database comparison (resource 1 vs resource 2), they will be a unique template containing the following columns:

  1. Reference pathway from resource 1. E.g., 'Akt signaling'.
  2. Proposed mapping pathways according to content similarity (Point 1 in the next section). The pathways (resource 2) are represented by an ordered list of top 10 most similar pathways together with the values of similarity (%). E.g., ['Akt signaling pathway'|90%, 'MAPK signaling'|50%]
  3. Proposed mapping pathways according to name similarity (Point 2 in the next section). The pathways (resource 2) are represented by an ordered list of the top 5 most similar pathways. E.g., ['Akt signaling pathway', 'Example signaling'...]
  4. Curator's proposed mapping. This column contains the mappings proposed by the curator. In case there is more than one pathway, the mappings need to be separated by pipes (|). E.g., "Akt signaling pathway equivalentTo Akt pathways|Akt signaling pathway isPartOf Akt family pathways")
  5. Curator explanation. Arguments behind the mapping, needed for the inter-curator agreement.

We have chosen to use the top of most similar pathways instead of presenting the curator with the full list due to the vast amount of overlapping pathways. If we present the curator with 100 pathways to analyze against one, the curator would need huge amount of time. Instead, the curator is presented with overlapping pathways, order by similarity and top 5 pathways with the most similar name based on string similarity. While the ranked pathway list based on similarity provides insight on pathways sharing genetic similarity, the other one gives suggestion on possible pathways based on the name. Furthermore, hyperlinks to the description were attached so that the curator can confirm that the focus and context of the pathways remain the same, even if the both have high similarity and name.

Inter-curator agreement

The final step of the curation process is to agree on the proposed mappings. Therefore, all three curators examined together one by one the mapping files and compare the differences, and subsequently agreeing on the final mappings.

Types of Mappings

ComPath curation interface allows for two types of mappings: equivalentTo and isPartOf.

  • equivalentTo.reflects that the pathway pair shares a high similarity. The guidelines for this mapping are:
    • The pair should share at least a certain degree of similarity (this is required). In other words, only pathways with at least 1 overlapping gene can be used for the mapping.
    • Same focus and biological function. A high concordance in the name between the pair is not required but it is helpful to identify this similarity. Curators have read the descriptions for each database to assess whether the focus of the pathway is the same or not.
  • isPartOf.This type of mapping represents a hierarchical relationship between the pathway 1 (child) and 2 (parent).
{% include "footer.html" %} {% endblock %}