Skip to content

Fix compile error, improve error message and docs#20

Open
roland-ewald wants to merge 1 commit intoPacificBiosciences:developfrom
roland-ewald:fix-compile-problem
Open

Fix compile error, improve error message and docs#20
roland-ewald wants to merge 1 commit intoPacificBiosciences:developfrom
roland-ewald:fix-compile-problem

Conversation

@roland-ewald
Copy link

This adds brief instructions on how to build pbmm2 (in particular, which flags need to be set to build it as an external user) to the README, and improves the error message that is printed when e.g. mm2helper->Align(...) is throwing an exception.

Two other points (raising them here as GitHub issues are not enabled for the project):

  • I've noticed that this builds version 1.13.1 (not 1.17.0); is there any way to build a newer version locally?

  • Sometimes it is preferable to fail early when the aligner throws exceptions (e.g. because of invalid BAM tags). Right now these errors are printed out but the program continues and returns the non-error code 0. Maybe we could add a --strict option to control this?

This adds brief instructions on how to build the software
(in particular, which flags need to be set to build it as an
external user) to the README, and improves the error message
that is printed when e.g. mm2helper->Align(...) is throwing an
exception.
@TedBrookings
Copy link

These changes are important to me. I have BAMs that seemingly align correctly in minimap2, do not produce any mapped alignments in pbmm2, and give the unhelpful error message ERROR with return code 0. The changes here would allow me to understand what problems with any there are with the BAM. I'm guessing it's an issue with one of the tags.

@roland-ewald
Copy link
Author

@TedBrookings Would it help you if I uploaded the binary? (You should of course not trust me enough to run this outside a container 🥲, but I think I still have it somewhere, compiled on Ubuntu 24.04).

@TedBrookings
Copy link

@roland-ewald Thanks for the offer! I have a local branch that was ready to build, and your PR contained the fixes needed to get that built successful, so a binary is not necessary.

That said, for the pipelines I'm running, I'm very wary of using a tool that can return success while incorrectly aligning no reads; and maintaining my own branch of pbmm2 is not viable. So absent a fix, I'm likely to migrate to using minimap2 with pacbio presets.

@armintoepfer
Copy link
Member

We are no longer providing the source code here, but only provide static binaries; it's noted in the release notes.

If you have problems with the tool and get ERROR, I would ask you to provide a minimal example, so we can fix it.

@TedBrookings
Copy link

There is no issues section in this repo, so I filed an issue in pbbioconda. The file I linked is large, although I I have a 1.5 GB subset BAM that recapitulates the error. I'm happy to provide this file to you if you have a place I can upload it.

@armintoepfer
Copy link
Member

https://github.com/PacificBiosciences/pbmm2?tab=readme-ov-file#binary-availability points you to "Support", which has an issues tab.

If you could break it down to a few reads that reproduce it, it would help.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Development

Successfully merging this pull request may close these issues.

3 participants