From the 25.09 release onwards, when a sequencing position crashes or is restarted, MinKNOW™ automatically detects and restores any recoverable POD5 files, moving them into a dedicated pod5_recovered directory within the experiment output folder.
Users are kept informed throughout the process: a system message appears in the MinKNOW interface confirming successful recovery and providing the location of the recovered files. If recovery is unsuccessful, a clear error message is displayed in the System Messages tab.

For MinKNOW versions from 24.11 until 25.05, raw read data can be left in the /data/queued_reads/complete_reads_xxxx directory (more details in this FAQ). Most of these data should be in the .pod5 file format and can be used directly.

Some files that MinKNOW was actively writing to may be in a .pod5.tmp format. They can be manually renamed to .pod5 and converted using the pod5_recover tool in the following manner:

pod5 recover broken.pod5

New files are written as siblings to the inputs with the _recovered.pod5 suffix.

On MinKNOW versions prior to 24.11, raw read data will be left in the /data/queued_reads/complete_reads_xxxx directory without being converted to pod5 files.

These raw data files can be converted into .pod5 files using a command line tool that is packaged within the MinKNOW suite:

1. Open up a terminal window
2. Navigate to the directory containing the recover_reads executable, by using the change directory command cd followed by the recover_reads executable location, usually in /opt/ont/minknow/bin for Linux users, and in /Applications/MinKNOW/bin for MacOS users. Newer MacOS users will find this located in /Applications/MinKNOW.app/Contents/Resources/bin/
3. Run the following command to recover your raw reads: ./recover_reads "location of your raw reads" --output-directory "location to output your recovered pod5s

Please Note: The complete_reads_xxx folder should be the one generated during the time of the sequencing run that ended improperly. An example command is shown below: ./recover_reads /data/queued_reads/complete_reads_xxx --output-directory /data/../output

Reads will be recovered to the --output-directory specified. The recovered files can be basecalled using post-run analysis in MinKNOW or Dorado.