I’m having problems with Kraken2 recently.
I run Kraken2 normally with the option to split classified. When starting with a fastqsanger.gz file, the output gives the same output format in the file type, but it’s actually an uncompressed file. At recently I could just compress the file and it appeared to be able to convert back to a compressed fastqsanger.gz file. But now this has changed, and Galaxy tells me it’s a binary file now and I’m unable to use it. Is there some way to get it back into a fastqsanger.gz format?
Hi @Jon_Colman
Ok, this is pretty interesting and I’d be very interested in taking a closer look. We are preparing for a new release of Galaxy – which makes this even more important to troubleshoot (and fix!) if I can reproduce it. Plus, we can probably offer a workaround.
It sounds like the tool had some small issue to start with, which you had your workaround for, then that became unstable. We can try to address all of it so you won’t need to do anything special at all.
Please generate and post back as a reply a link to a history that demonstrates the problem. That can be a simplified history, or the one you are working in now. I’m most curious about how this job was set up, and the versions of the tools/databases in use. The shared history will include all those technical details.
- How to get faster help with your question
- Directly link to the FAQ → FAQ: Sharing your History
Thanks for reporting the issue, and I’ll watch for your reply. 
So here is how I got around the issue. It only happened on my fungi output reads for some reason. I transferred the files to my galaxy eu by link, where they arrived as fastqsanger.gz. I wasn’t able to load back to galaxy via link, so I just downloaded and uploaded and it all seemed to be ok.
Ok, I’m glad you worked this out!
Do you mean moving the files from the EU server “back to” a different server or an external file location?
If yes, next time, to transfer files from the UseGalaxy.eu server to somewhere else, including another public Galaxy server’s Upload tool, you will need to set the history with the dataset to a shared state first (the “share via link” toggle). Then the links on datasets will also “work” and be able to send the data.
This is just a quirk of how the GDPR implementation functions. Pease give it a try if you need to move more data around, since it will avoid the download/upload step (if I understanding what is going on!). 
Now, if there is something going on with the Kraken2 outputs themselves that is still odd, I would still be interested in looking at an example. You shouldn’t need to move data between servers to access any of those outputs! Please let us know if we can help more with that part if not solved (I wasn’t sure from your comments so far). Thanks!