dgrun / varid_analysis Goto Github PK

Hi,

I met a problem about the large dataset. I have more than 30k cells in my data. I tried to follow the manual of raceid (https://cran.r-project.org/web/packages/RaceID/vignettes/RaceID.html). According to this, I didn't use the normal pipeline ("sc <- clustexp(sc)") and tried the pipeline for large datasets. Followed is my code:

###############
sc <- SCseq(x)
sc <- filterdata(sc,mintotal=1000,CGenes=rownames(x)[grep("^(mt|Rp(l|s)|Gm\d)",rownames(x))])
expData <- getExpData(sc)
res <- pruneKnn(expData,no_cores=5)
cl <- graphCluster(res,pvalue=0.01)
probs <- transitionProbs(res,cl)
noise <- compNoise(expData,res,regNB=FALSE,pvalue=0.01,no_cores=5)
################
When I runned the command: pruneKnn, it reported an warning:
##############
Warning in irlba(A = t(z), nv = pcaComp) :
You're computing too large a percentage of total singular values, use a standard svd instead.
Warning in irlba(A = t(z), nv = pcaComp) :
did not converge--results might be invalid!; try increasing work or maxit
###############
I didn't know what's happend of this warning.

Meanwhile, when I tried the compNoise, it reported an error:
#################
Error in lm.fit(x, y, offset = offset, singular.ok = singular.ok, ...) :
0 (non-NA) cases
#################
It seems there are some NA in my data. But I don't know how to deal with it?

Could you give me a hand? BTW, whether I did right or not to deal with the large datasets using the pipeline?

Thank you!
Best,
Garen