All of the older Tuxedo tools are in a deprecated state but some still work. This includes
Cuff* tools, and
Tophat for Illumia, in particular, should definitely be avoided.
Cuffdiff does have a known issue with the SQLlite output option (the input for
cummeRbund) and that will not be fixed. There are probably other issues with that tool and/or other deprecated tools.
We strongly recommend updating to use current/supported (eg: maintained) differential expression tools/methods. Tutorial links below, plus an FAQ that covers the various usage issues that can come up and be addressed – many will apply to the tools/methods you are using now – as well as other tools/workflows. It is a list of summarized help for the most common usage errors.
That said, you could try at least one more rerun. Sometimes the particular error you got comes up due to transient cluster issues. Even if you reran already, try one more time unless you discover that there was an input problem first.
If it fails again, then check your inputs more carefully. If those are correct, you probably will need to move to use the current tools/methods for scientifically correct results, even if a deprecated tool does not actually error. We will not be addressing fixes for any deprecated tools and can offer only limited support if you decide to use them anyway for some reason.
The FAQ that includes both DE troubleshooting tips plus the current DE tutorial links: https://galaxyproject.org/support >> FAQ >> Extended Help for Differential Expression Analysis Tools
Note: If you are using
HISAT2 for mapping or decide to use it going forward, a specific setting to format the output for
Cufflinks is required. The same
BAM result cannot be used with both
Stringtie. Find the setting on the tool form under Advanced Options > Spliced alignment options > Transcriptome assembly reporting. Again, it is strongly recommended to NOT use
Tophat or Tophat2 – both can produce unreliable scientific results or metadata problems that can be difficult to detect, even if the mapping job did not actually error for technical reasons. For metadata problems, sometimes you’ll need to expand the mapping dataset to see the warning in a green, putatively “successful” job. A link is given to “redetect” the metadata. Sometimes it works, sometimes not. And even if it works, there could still be content issues.
More troubleshooting help, summarized with more tips/links for common usage issues and solutions:
Hope this advice helps!