Watching SHOC2 evidences for both cancer and neoplasm pathologies, I’m surprised to get a bigger score for onthology cancer “0.60793085” than for neoplasm “0.5404505” for the class of data “Pathway & systems biology”.
Actually, if you have a look at the evidences for the association SHOC2/neoplasm, you’ll find two of them (Project Score and Reactome) :
The first thing to note is that an association with more pieces of evidence is not necessarily a higher scoring association.
For example, the scores for certain data sources will be downweighted if we consider the evidence to be less relevant (e.g. from text mining rather than manually curated). In the case of Project Score, we use the priority score divided by 100, whereas manually curated evidence in Reactome has a score of 1.
Once we have scores for each data source (e.g. Project Score, Reactome), we calculate the data type score (in this case, for Pathways and Systems Biology), using a harmonic sum.
