1.9.1 Linux write manifest error

[karlssoc]

Thanks for the updates and for providing the community with free exceptional software.

In my testing of diann-linux v1.9.1 it always crashes when writing the manifest file and no manifest file is written.

I tested with:

• Thermo .raw files converted to indexed .mzML (ThermoRawFileParser v1.4.4) or to .dia (Windows dia-nn v1.9.0)
• library-free or with speclib
• using singularity container (debian:latest) or with native enviroment (Ubuntu 22.04.3 LTS)
• with and without the mimalloc library

The same input files works fine with diann-linux v1.8.1.

Below is the the error I get, same in all cases.

[3:08] Protein group levels matrix (1% precursor FDR and protein group FDR) saved to test_diann_1.9.1/report.pg_matrix.tsv. [3:08] Saving gene group levels matrix [3:08] Gene groups levels matrix (1% precursor FDR and protein group FDR) saved to test_diann_1.9.1/report.gg_matrix.tsv. [3:08] Saving unique genes levels matrix [3:08] Unique genes levels matrix (1% precursor FDR and protein group FDR) saved to test_diann_1.9.1/report.unique_genes_matrix.tsv. [3:08] Manifest saved to test_diann_1.9.1/report.manifest.txt terminate called after throwing an instance of 'std::length_error' what(): basic_string::_M_create

[vdemichev]

Thank you for letting me know, will take a look what's up with it. A workaround for now would be to not use matrix generation on Linux.

[karlssoc]

Wow, thanks for the quick reply. I can confirm that removing matrix generation indeed works.

[SnowcastR]

I also had the same problem for Linux 1.9.1 and removing --matrices does work.

[vdemichev]

Sorry, can you please clarify exactly what you are doing and what expected information is missing from the output? Sample information is in the 'Run' column of the .parquet report. (Note that the spectral library is also saved in .parquet format by 1.9.1, i.e. need not to confuse the two parquet files :))

[ReddysLab]

Apologies - yes, we got the wrong parquet file. Thanks!

[lyons89]

I can also confirm this issue. 1.8.1 on linux runs great, but using 1.9.1 with the same .mzML files triggers the same error that OP showed. Removing --matrices from config file fixes it.

[Aaron-Amunix]

Hi @vdemichev we are having this same issue, are you working on a fix for this? We are having some trouble extracting the data from the parquet files in our pipeline. If there is a fix planned soon? If not we'll keep trying to work through this.

[Mnago]

Hi @vdemichev We are having this same issue. We are using the DIA-NN v1.9.1 on a Linux system to analyze DIA files and encountered the following error: 'terminate called after throwing an instance of 'std::length_error' what(): basic_string::_M_create'. I have also attached the search database parameters. How can I resolve this issue?