• Acknowledgment & funding
• Frequently asked questions
• How to use
• The people behind GORBI
• License

Frequently Asked Questions (FAQ)

Why are there predictions only for prokaryotes?

GORBI is based on phylogenetic profiling, a method where the patterns of gene presence/absence across genomes is used to predict gene function. This works better for Bacteria/Archaea than for eukaryotes, see eg Škunca & Dessimoz (2015, PLOS One) and references therein.

How do I cite GORBI?

We hope you found GORBI useful in you work! If so, please cite: Škunca et al. Phyletic profiling with cliques of orthologs is enhanced by signatures of paralogy relationships. Open access from PLOS Comp Biol 2013.

How did you validate the predictions?

We validated the predictions from the original GORBI-2013 version (from our PLOS Comp Biol paper) by performing experiments on 38 E. coli knock-out mutants, please see here.  

The updated, GORBI-2015 version was independantly validated by competing in the CAFA2 challenge for automated function prediction methods. The results of CAFA2 will be available soon. 

What is Precision?

Precision (Pr) stands for the fraction of correctly predicted examples we expect out of all the predictions made. It is equivalent to 1-FDR (false discovery rate).

For example, Pr associated with the E. coli gene yfgI for "response to DNA damage stimulus" was 62%; for "translation" and "peptidoglycan-based cell wall biogenesis" it was lower than 1%. We would therefore predict this gene to be involved in "response to DNA damage stimulus" with a probability of being a false positive of 38% (=100%-62%). For the other two GO terms the probability of being a false positive would be over 99%: we thus infer that these are unlikely functions for this gene.

I need a specific output with custom parameters. How can I get this?

This might happen if, for example, you need a more permissive parameter for Precision (the website holds only those predictions where Precision is higher than 45%), if you expect the query to return too many results and the web-based query would take too long, or you want specific GO terms in the results.

Please send an email to gorbi AT irb DOT hr with your request and we will do our best to help you.

What to do if there are no results?

Tips to broaden the search:

* Try with a less stringent Precision cutoff.
* Select the parent Gene Ontology term of your query in the Gene Ontology tree.
* Would a different strain of the query species be appropriate? Try browsing the Taxonomy tree, starting from a higher rank of the query species.

If the above fails, contact the authors to obtain less specific predictions: the Web site only presents the predictions that have Precision higher than 0.45