Hi Nate
Explicitly specifying the files names make the upload to work the way I expected it should. I am planning to discuss few of such changes with Anton/Greg sometime next month specially in context of NGS related data.
Also when the data is exported from library to history for analysis, is is actually copied ? I see that part taking considerable time.
Thanks,
-Abhi
On Thu, Oct 8, 2009 at 3:37 PM, Nate Coraor
<nate@bx.psu.edu> wrote:
Abhishek Pratap wrote:
Based on the testing of the new feature I am not sure if it working as expected. I gave galaxy a path to GERALD folder (which is a analysis folder generated by Illumina pipeline) and selected file type to be just "solexafastq". It took about two hours for the process to be completed. Somehow it went through each file and uploaded everything in the library. Ideally it should look up only the relevant fastq files and upload total of 8 files (1 / lane ) present in the folder. This process should be very fast as it would just record the path.
The file type selector doesn't determine what type of files to search for - partially because it predates any sort of multiple upload feature, and partially because it's not always possible to trust a file's extension (if it even has a recognizable extension). If such a feature would be useful, a feature request can certainly be created for it.
If you only need 8 files out of a large directory of files, it'd be best to simply paste the path to those 8 files in the upload box.
Also, be sure you're checking "No" to the "Copy data into Galaxy?" box to ensure data is only "linked" and not copied.
--nate