4.4 years ago by
United States
Hello,
Uploaded (browsing or FTP or "Get Data -> EBI SRA") datasets originally in .gz format will uncompress automatically when loaded into a history. This is true on the Main public Galaxy server at http://usegalaxy.org, Cloudman instances http://usegalaxy.org/cloud, and local Galaxy instances by default (when FTP is enabled with the rest at basic configuration) http://getgalaxy.org - plus many Public Galaxy instances https://wiki.galaxyproject.org/PublicGalaxyServers (unless the administrator disabled the option - direct contact with their support would be the best way to trouble shoot, but let we can help you to make contact/clarify details if you need assistance).
UPDATE: tar.gz should not be an issue from SRA through routine methods, but please know that tar archives will uncompress just the first dataset, discarding the rest, so are not recommended. Uncompress the archive first locally, then load dataset indivudually (recompressing each as .gz will speed loading).
I suspect that you are noting the dataset name (will often still have the .gz appended to this attribute), but what you want to examine is the "datatype" metadata. You'll be able to visualize this in the UI in the dataset's box - is clearly labeled. For new uploads, unless format "autodetect" was not selected, the "datatype" will most likely simply be ".fastq".
To work with this data further in Galaxy, determining the correct quality scaling and assigning a more specific datatype that describes the quality score scaling (rescaled with the tool "FASTQ Groomer" if needed), is the next step. This wiki section has instructions, including a screencast example. This is incredibly important (essential!) to get correct at the very start of an analysis run for valid results.
https://wiki.galaxyproject.org/Support#FASTQ_Datatype_QA
The accession from the example above can be used with the tool "Get Data -> EBI SRA" as a search to locate the exact full dataset. As is the case with the majority, if not all, SRA data obtained this way, there are two options for import into Galaxy. The table for an experiment will have two import obtains: "submitted" which is the original author data (raw) and the "processed" that has undergone manipulations by SRA to scale the quality scores to Fastq Sanger Phred with an ASCII offset of +333 - otherwise known as the datatype label ".fastqsanger" in Galaxy. Use the "processed" to quickly move forward with analysis. Run FastQC to confirm (also highly recommended for other basic metrics- e.g. QC for content). Most tools require this "datatype" assignment be made as a starting point for correct data interpretation and some will not even recognize a .fastq dataset as valid input without it.
I provided a bit more information that you asked for, but all is interrelated, and overlooking these early QA steps are a quite common cause of issues downstream. So, I tend to share whenever I can! Please make use of what is helpful for your case, and hopefully it will aid others reading - in a similar situation or just following new posts.
If you continue to have what you suspect to be loaded uncompression problems, please explain a bit more about where you are working and the exact steps you have done so far. Include the metadata assignments for the dataset. We can work from there to resolve any outstanding issues.
Cheers! Jen, Galaxy team
