Hello @Atefeh_Bahmei
In general, you will want to load data into Galaxy using all default settings with the Upload tool. This performs a data content check and a format datatype will be assigned that matches the file content. Tools then interpret this datatype for manipulations (including upstream file conversion steps, as needed).
During this data loading process, data in gtf.gz format will be automatically converted to gtf format and you should see that assigned to the dataset. If you don’t see that format assigned, then there is a format deviation that will need to be addressed. Sometimes you can do this in Galaxy – it usually involves simply removing the headers.
If you tried something else originally, loading this way may be enough if directly uncompressing is not working for some reason. My guess is that the file has an incorrect datatype currently and that is preventing the “uncompress” function from working correctly. The file may already be uncompressed but labeled as compressed?
If nothing else works: try to uncompress the file on your own computer, then load the data in an uncompressed format to Galaxy. This confirms that the file is intact (can be uncompressed) and gets around any special gz compression types that may be specific to a particular operating system.
Why the mapping job didn’t complain about the file is hard to guess about. But I can let you know that the underlying mapping tools will “ignore” data lines in a reference annotation file that are not paired up with the reference genome. This is intentional behavior but it could lead to the entire file being not used! If this is a concern, consider rerunning with the same version of the uncompressed annotation that you will be using for these later downstream steps.
Please give this a try and we can follow up more until this is resolved for you! 