[BUG] DeseqDataSet.deseq2() still prints when quiet=True is set

[yihming]

Describe the bug

My following code

from pydeseq2.dds import DeseqDataSet
from pydeseq2.default_inference import DefaultInference

inference = DefaultInference(n_cpus=1)

dds = DeseqDataSet(counts=..., metadata=..., design_factors=..., inference=inference, quiet=True)
dds.deseq2()

still prints the following 3 sentences when quiet=True, while from the source code they should be controlled by quiet parameter:

Fitting dispersions...
... done in 0.20 seconds.

Fitting MAP dispersions...
... done in 0.22 seconds.

Fitting LFCs...
... done in 0.17 seconds.

To Reproduce

This case happens for some data, not all. However, my case uses internal data, so I may have to do some anonymization work before sharing it with you.

Expected behavior

These sentences should not be printed if quiet=True is set.

Screenshots

Already shown above.

Desktop (please complete the following information):

• OS: Ubuntu 22.04.4 LTS
• Version: pydeseq2 v0.4.9

Additional context
Add any other context about the problem here.

[umarteauowkin]

Hi ! Thanks for reporting. I haven't been able to reproduce yet. Could you assess if the self.quiet attribute has been changed during the pipeline ? (typically by debugging before printing one of these three sentences). It would be a great help. Thanks !

[yihming]

Hi @umarteauowkin ,

In the attachment is a test file which triggers the issue at my side. The counts is count_df.csv, and metadata is metadata.csv. The design factor is condition, and contrast is ["condition", "A", "B"].

count_df.csv
metadata.csv

[yihming]

And just let you know that all the 3 places show that self.quiet is False. However, I cannot find any other places where self.quiet is changed in dds.py file.

However, if I print self.quiet before and after each step in deseq2 function, it's all True...

[umarteauowkin]

I m quite puzzled here. I will take a look (but maybe more in July as this is not too blocking). Thanks for the minimal example !