ValueError: The first guess on the deviance function returned a nan. This could be a boundary problem and should be reported.

[mohammedali452]

I get the issue from the title when running dds.deseq2() after initializing dds as a DeseqDataSet. Any suggestions on how to fix this error?

[BorisMuzellec]

Hi @mohammedali452, could you provide a MWE to reproduce this behaviour?

[mohammedali452]

dds = DeseqDataSet(counts=test_gene_expression_df, metadata=test_metadata_df,design_factors="Cluster") dds.deseq2()
Let me know if you need to look at the counts and metadata dataframes.

[BorisMuzellec]

Let me know if you need to look at the counts and metadata dataframes.

Yes, if possible

[mohammedali452]

Count Data: https://drive.google.com/file/d/1ooydVzIGxRUo8HsX7GbjdoBpDJpO-28D/view?usp=sharing
Metadata Data: https://drive.google.com/file/d/1fwptdrom4syILtlZnSUINOq86Yhh8GfP/view?usp=sharing

[BorisMuzellec]

@mohammedali452 are you performing any gene / sample filtering step?

Your data is very sparse: it's 98% zeroes and more than half of your genes have full zero counts. I don't think DESeq2/ PyDESeq2 can handle this level of sparsity.

Is this single-cell data?