Frequently Asked Questions

The web application is not behaving as expected; what is wrong?

We test with the latest versions of Chrome, Safari, and Firefox, but primarily develop and test with Chrome, available at https://www.google.com/chrome.

You can send bug reports to pathfx_support@googlegroups.com; please include your OS and browser. In the meantime, please try using Google Chrome and see if that works for you.

What if all of my results files are empty?

When all results files are empty, it means that PathFX did not have sufficient protein targets with which to create a network. This can result if no drug-binding genes are provided (either by the user or by the stored set from DrugBank) or if the drug-binding genes are not in our interaction network. Please read the log file for a more detailed algorithm status.

I provided all of the correct inputs but am getting empty results files.

If the protein targets are not within the stored interactome, PathFX will also be unable to draw a pathway. Occasionally, one or some of the protein targets are not within the interactome. In these cases, PathFX skips these targets and completes analysis with the remaining targets. Please read the log file for a more detailed algorithm status.

Why does phenotype clustering take so long?

Unfortunately, we don't have the capacity to parallelize the comparisons per query or precompute all possible combinations ahead of new search queries. We measure compute time against number of input phenotypes and found that on average, clustering of 50 phenotypes could occur in about 24 hours.

Should I run phenotype clustering?

Phenotype clustering provides a high-level view of the top 50 phenotypes associated with a drug network. For some users, reading the association table is sufficient. Others prefer to have a visual, dendrogram relating terms to each other to look for similarity among associated phenotypes.