ProteInOn is a web tool focused on calculating GO-based protein semantic similarity. It features a stepwise query selection menu, which together with the possibility of selecting results as input for new queries, makes it flexible and customizable. It also incorporates data on protein interactions, allowing for comparative studies between protein similarity and interactions.
The tool can be used to compute semantic similarity between proteins or GO terms, using several different measures: the three “classic” term similarity measures (Resnik’s, Lin’s and Jiang and Conrath’s) with or without the GraSM approach and using Best-Match Average combination of term similarities; as well as the novel simGIC.
It can also be used to find GO terms assigned to one or more proteins or GO terms representative of a set of proteins. A score is used to measure the representativeness of a term for a set of proteins, based on the number of proteins the term is annotated to and its probability of annotation.
Finally, it can be used to find proteins that interact with one or more other proteins, which enables the integration of knowledge on protein interactions with functional and processual knowledge: proteins that share a group of interactors should have a similar molecular function whereas a group of interacting proteins should be involved in similar biological processes.