2.5 years ago by
The option for downloading the genome into the history is intended to reduce load. Download it once, then reuse. However, this usage seems to be problematic (feedback from our team is pending). I suspect there is an issue with the dbkey assignment between the inputs (meaning, the database assignments for "dbkey" are a mismatch). SnpEff genomes include incremental versions in the key, while native genomes at http://usegalaxy.org do not plus are often named in a different way (using UCSC identifiers, etc). It may be possible to create a custom reference genome "build" that uses the exact same dbkey as SnpEff (database attribute) that would allow this option to work, but that has not been tested by me and seems tedious for large genomes. It could also trigger a memory problem, since that custom reference genome would need to be used for all steps in the analysis (not just the SnpEff annotation step).
Using the option "Named on demand" does seem to functions in small tests. Although I should mention that certain tool options on the tool form are passed to the command-line in deprecated format in many cases. This issue can be tracked here (and may not be the root issue): https://github.com/jennaj/support-known-issues/wiki
I suggest using the "Named on demand" option when working on http://usegalaxy.org. If working on your own local/cloud, then the native genome indexes could be created in a way that the dbkey is a match for the SnpEff genome dbkeys, and tested. Problems can be reported to the tool authors through the Tool Shed (http://usegalaxy.org/toolshed) or in Github.
If our team has more feedback, we will post an update.
Sorry for the confusion in usage, Jen, Galaxy team