Richard Harris

Hi @caspargross, Your duplex command is malformed: ``` bash dorado duplex ... --modified-bases-models ``` This arrangement results in the simplex model being consumed by the `--modified-bases-models` argument which accepts multiple...

Apologies for not replying sooner. I'm pleased to see your issue is resolved. For future reference, generally we recommend reducing batchsize over chunksize as changing batch size will have no...

Hi @janpb, We've seen this issue in a couple of places now - and we're investigating. Thanks for the detailed report.

Hi @ilivyatan You're just missing the required `--output-dir` argument. The readme could be clearer on this for sure: When reading from an input folder, dorado aligner also supports emitting aligned...

https://dorado-docs.readthedocs.io/en/latest/basecaller/alignment/#writing-to-an-output-directory

@pre-mRNA Can you also try `--chunksize 9996`? This may further reduce memory consumption.

Sorry to hear you're still having difficulties. We're working on improving the performance of the new sup model architecture and modbase models to improve the situation in the future.

@caspargross, Does this issue persist in 0.9.0+ which has a new modbase model architecture which improved performance? Kind regards, Rich

Hi @jainronit, This looks like it could be an issue with the fast5 file. Can you try the script [here](https://github.com/nanoporetech/pod5-file-format/issues/97#issue-2037612435) to find issues in the fast5 file?

Do you have any other log files other than `...main-pod5.log`? There should be another one for the conversion sub-process which might be insightful. Kind regards, Rich