Stuck at Fitting MAP dispersions... ... done in 0.54 seconds.

[xin4396]

When doing DEG analysis, I declined the number of tested cell from 20,000 to 4000, the dds.fit_LFC() wont start:

🔬 分析 TAK-901 vs Control (整体数据)

🔬 开始分析: TAK-901 vs DMSO_TF
分析名称: Overall_TAK-901_vs_Control
原始样本数 - TAK-901: 271, DMSO_TF: 1776
分析样本数 - TAK-901: 271, DMSO_TF: 1030
过滤后基因数: 4767
⚙️ 运行PyDESeq2分析...
Fitting size factors...
/home/clab/anaconda3/envs/geneformer/lib/python3.10/site-packages/pydeseq2/dds.py:532: UserWarning: Every gene contains at least one zero, cannot compute log geometric means. Switching to iterative mode.
self.fit_size_factors(
Using None as control genes, passed at DeseqDataSet initialization
Fitting dispersions...
... done in 0.63 seconds.

Fitting MAP dispersions...
... done in 0.61 seconds. keep stucking :(

[xin4396]

I found that though deseq2 has printed "Fitting MAP dispersions...
... done in 0.61 seconds", the true fit_MAP_dispersions() never ends, since it will print "MAP finished":
dds.fit_MAP_dispersions()
dds.var["MAP_dispersions"]
dds.var["dispersions"]
print('MAP finished')

    dds.fit_LFC()
    dds.varm["LFC"] = dds.compute_LFC()

[BorisMuzellec]

Hi @3220111903bit, sorry for the late reply.

If the full trace you get this one?

Fitting size factors...
/home/clab/anaconda3/envs/geneformer/lib/python3.10/site-packages/pydeseq2/dds.py:532: UserWarning: Every gene contains at least one zero, cannot compute log geometric means. Switching to iterative mode.
self.fit_size_factors(
Using None as control genes, passed at DeseqDataSet initialization
Fitting dispersions...
... done in 0.63 seconds.

Fitting MAP dispersions...
... done in 0.61 seconds.

If so I think that the size factor step isn't actually finished and that the issue is somewhere in self._fit_iterate_size_factors (which launches its own dispersion estimation).

I'm not sure what's wrong there though.

For how long did you keep the code running (stuck)?