Error in checkna(X) : Missing values found in 'X' from mapQuery()

[qicheng-ma]

Dear Symphony Team,

I got the following error message when I map the my query to my own reference.

I appreciate your help:

query = mapQuery(query_exp , # query gene expression (genes x cells)

• query_metadata , # querymetadata(cellsxattributes)
• reference , # Symphonyreferenceobject
• do_normalize=TRUE , ## performlog(CP10k+1)normalizationonqueryperformlog(CP10k+1)normalizationonquery
• do_umap=TRUE # projectquerycellsintoreferenceUMAP
• )
  Normalizing
  Scaling and synchronizing query gene expression
  Found 3281 out of 3281 reference variable genes in query dataset
  Project query cells using reference gene loadings
  Clustering query cells to reference centroids
  Correcting query batch effects
  UMAP
  Error in checkna(X) : Missing values found in 'X'
  Calls: mapQuery -> -> checkna
  In addition: Warning message:
  In log(1 + A@x) : NaNs produced
  Execution halted

[joycekang]

Hi Qicheng,

Thanks for the message. Based on the output you listed, it looks like the mapQuery function is able to make it to the UMAP step before producing this error? Would you be able to try it with do_umap = FALSE and output the query object str(query) to check whether there are any NA's present in any of the internal data structures?

[TimoTrefzer]

Hi,

I actually got the same error, when using mapQuery. And as you suggested, running with do_umap=FALSE lets the function run through. But indeed somewhere NA's seem to be introduced, as in all query$Z ... there are only NA values.
Do you now what could be the most likely cause?

Thanks a lot!

[jyang635]

Hi all,

Just in case anyone gets this error in the future: the issue in my case was caused by the standard deviation with 0s stored in the ref_obj$vargenes_means_sds, which introduced NaN values during scaling.

My solution was to remove the genes with 0 standard deviations from the variable gene list before building the reference.