Page Comparison

...

1. Ensure your experiment and data type is accepted by the GEO database. See here and here for more details.
2. Create an NCBI account if you do not already have one.
3. Collect the files necessary for the GEO submission process - metadata spreadsheet, raw data, and processed data files. These files should all be on the O2 cluster.
   1. To aid in transfer times, you can compress your raw data files with either bzip2 or gzip.
   2. You should include md5 checksums for your raw data files (to ensure the in your metadata spreadsheet; these will be used to identify if any files were not corrupted or transferred incompletely). You can calculate these checksums with the md5sum command on the O2 cluster.
4. Create a folder named with your NCBI username, and move all of the files you want to submit to GEO there.
5. If you will transfer 1TB or more of data, contact GEO before you do so. See here.
6. Transfer the data to GEO using the transfer cluster: transfer.rc.hms.harvard.edu 
   1. Your username and password for the transfer cluster are the same credentials you use to log in to O2 (eCommons ID and password). Your username must be in lowercase.
   2. You can use lftp to connect to the NCBI FTP server. To obtain the appropriate lftp command, log into GEO and navigate to this page. The command will be listed under "Uploading your submission" > "FTP instructions" > "Linux/Unix" > "Here is a typical 'lftp' session". The command will be in the form of: lftp ftp://username:password@server
   3. Once you're connected to the server, you can transfer your data like so: mirror -R GEOusernameNCBIaccount_directory
7. Send an email to NCBI when your transfer has completed. More information can be found here.

...