Issue Using MIBiG with BiG-SCAPE on Galaxy

[Shreyash_04]

Dear Galaxy Support Team,

I hope you are doing well.

I am experiencing persistent issues while running BiG-SCAPE on Galaxy, specifically related to the inclusion of the MIBiG reference dataset.

MIBiG issue

Whenever I enable the MIBiG option (e.g., --mibig21), the job consistently fails with a read-only filesystem error:

OSError: [Errno 30] Read-only file system:
'/usr/local/lib/python3.7/site-packages/bigscape/Annotated_MIBiG_reference/MIBiG_2.1_final'

As a result, I have not been able to run any analysis with MIBiG included. (Same issue with 3.1 version database)

---

Without MIBiG

When I run BiG-SCAPE without MIBiG, the tool works correctly:

• I tested inputs as:
  • a collection of region BGC files, and
  • a single combined file (This failed)
• The analysis completed successfully, indicating that my input files are valid

---

Summary of the problem

• Input files → working
• BiG-SCAPE (without MIBiG) → working
• Adding MIBiG → always fails

---

Questions

I would appreciate your guidance on the following:

1. Is there a supported way to run BiG-SCAPE with MIBiG enabled on Galaxy?
2. Is there a pre-configured/shared MIBiG dataset available to avoid this issue?
3. Could the tool be adjusted so that MIBiG is extracted or accessed from a user-writable directory instead of the system path?

Including MIBiG is important for my analysis, so having this functionality working would be very helpful.

Thank you very much for your support.

Best regards,
Shreyash

[jennaj]

Hello @Shreyash_04

The issue is likely with the collection input or the way the input file is named. We can help to review if you want to share back the history. This will allow us to troubleshoot any issues with the tool itself and get those reported, too!

And, just a disclaimer about the status of this tool (still a “beta” implementation).

@jennaj

The Galaxy implementation is still under experimentation, and only available at a single server

And an example about the collection/file naming was in this topic. It also has some examples if you’d like to troubleshoot more on your own first!

@Elia_Soriato

Thank you very much! The issue was indeed related to the collection’s naming. After changing it without underscores, everything worked perfectly. Thanks again!

That said, we did go through this a bit in your original question, but we didn’t see the final result yet. Is issue is different now?

@Shreyash_04

Thanks a lot for the solution. It worked perfectly. I tested it with the small dataset as well as with the larger files. It works well and has produced satisfactory results.

Let’s start there!

[Shreyash_04]

Hi Jennifer,

The issue is different than before. This time the problem is with the MiBiG dataset. The tool is somehow unable to obtain the MiBiG database. The run works well for clustering and analysis of BGCs itself without considering MiBiG BGCs.

Best regards

Shreyash

[jennaj]

Hi @Shreyash_04

Thanks for sharing the details! I was able to confirm the issue with all tool/database version combinations. I suspect the new job container is the problem, maybe with how it is being used.

I’ve opened a ticket against the EU server to get this started:

• Correction request bigscape: "Read-only file system" reported when a mibig database is used · Issue #2236 · usegalaxy-eu/infrastructure-playbook · GitHub

You can also send in a bug report from one of the red error datasets to given them another example.

I don’t have a workaround since the tool is only hosted at this server, and if the root issue is the container, the tool would probably fail no matter where it run.

Let’s see what the EU team thinks!