The tissue associations are derived from manually curated knowledge in UniProtKB and via automatic text mining of the biomedical literature, which has not been manually verified. The confidence of each association is signified by stars, where ★★★★★ is the highest confidence and ★☆☆☆☆ is the lowest. Download files from earlier versions are archived on figshare.
Each tissue–gene association is based on a text-mining score, which is proportional to 1) the absolute number of comentionings and 2) the ratio of observed to expected comentionings (i.e. the enrichment). These scores are normalized to z-scores by comparing them to a random background. This is represented by stars, each star corresponding to two standard deviations above the mean of the background distribution.
Developed by Alberto Santos, Oana Palasca, Christian Stolte, Kalliopi Tsafou, Sune Frankild, Janos Binder, Sean O'Donoghue, Jan Gorodkin, and Lars Juhl Jensen from the Novo Nordisk Foundation Center for Protein Research, Center for non-coding RNA in Technology and Health, and the Commonwealth Scientific and Industrial Research Organisation (CSIRO).
