No weights after zinbwave()

[ktroule]

Hi

I'm following the zinbwave tutorial with my own data. I want to perform a DEG analysis, but I do not get the weights for the counts matrix. summ.exp_norm is just a SummarizedExperiment object with a raw count matrix in assay

zinb <- zinbFit(summ.exp_norm, K=2, epsilon=1000)`

> zinb
Object of class ZinbModel.
7494 samples; 635 genes.
1 sample-level covariate(s) (mu); 1 sample-level covariate(s) (pi);
1 gene-level covariate(s) (mu); 1 gene-level covariate(s) (pi);
2 latent factor(s).

summ.exp_norm <- zinbwave(summ.exp, fitted_model=zinb,K=2, X="~Patient", epsilon=1000,
+                             normalizedValues=TRUE, residuals = TRUE)

> summ.exp_norm
class: SingleCellExperiment 
dim: 635 7494 
metadata(0):
assays(3): counts normalizedValues residuals
rownames(635): TXNIP CCL5 ... PPP3R1 HSPA6
rowData names(0):
colnames(7494): Cell_1 Cell_2 ... Cell_7493 Cell_7494
colData names(3): Cell.id Cell.type Patient
reducedDimNames(1): zinbwave
spikeNames(0):

I've tried different combinations just in case.

I know zimbwave performs normalization only for visual ization purposes, but how does it deals with batch effect (2 conditions but from several patients) or normalization of the data.

thanks for your help

[drisso]

Hi Kevin,

which version of zinbwave are you using? Note that the function zinbwave computes observational weights only since version 1.1.5, so if you have an older version, you may need to either set observationalWeights = TRUE or to use the stand-alone function computeObservationalWeights depending on how old your version is.

In any case, I recommend that you update to the latest version of the package.

The details on how zinbwave deals with batch effects are in the paper, available here: https://www.nature.com/articles/s41467-017-02554-5

Please, let me know if there are specific points that are unclear.

[ktroule]

Thanks. zinbwave version is older. I'll read again the paper.