Precision and recall of pubmed2ensembl data sources relative to the BioCreAtIvE I and II Gene Normalization data sets.

Gene-PMID pairs found in both a pubmed2ensembl data source and a BioCreAtIvE I and II Gene Normalization data sets were considered true positives (TP). If a gene-PMID pair was present in a pubmed2ensembl data source but not in a BioCreAtIvE data set, it was considered a false positive (FP). Conversely, if a gene-PMID pair was not present in a pubmed2ensembl data source but was present in a BioCreAtIvE data set, it was considered a false negative (FN). Precision was calculated as TP/(TP+FP); Recall was calculated as TP/(TP+FN). Rows labeled with Total indicated the total number of PMIDs and gene-PMID pairs in the BioCreAtIvE evaluation data set for each species. Values with division by zero are labeled “N/A”.