dtsoucas / dwls Goto Github PK

Dear guys,

Is it possible to get the cell type expression values in bulkRNAseq after DWLS?

Say, there are 10 cell types in the bulkRNAseq data (deconvolution by scRNAseq, which has 10 cell types); is it possible to only extract the cell type 5 expression in the bulkRNAseq data?

Hi

I am using DWLS in my data, but I found the predicted proportions is negative, do you know why this happened?

Dear DWLS developer,

I found your method appealing and would like to apply it on our personalized data. I am currently stuck at running solveDampenedWLS() with part of the samples(while others works fine). I get the Error:

Error in solve.QP(D, d, A, bzero): constraints are inconsistent, no solution!
Traceback:
1. solveDampenedWLS(S, B)
2. solveOLSInternal(S, B)
3. solve.QP(D, d, A, bzero)
4. stop("constraints are inconsistent, no solution!")

Can you suggest a way I can fix this issue?

Thank you!

I was trying to install the DWLS package in R via devtools::install_bitbucket("yuanlab/dwls", ref="master"). But the installation gives this error Error: object 'zlm.SingleCellAssay' is not exported by 'namespace:MAST'.

The available versions of MAST do not export SingleCellAssay. So, this error cannot be resolved by installing any of the MAST package versions either.

Below you can find a screenshot of the error generated in R. Could you please tell a way to resolve this?
link
I hope to get a reply from you at the earliest.

Thank you.

Dear Daphne Tsoucas,

I am trying to run the example mentioned in the vignette with the files "dataSC.RData","dataBulk.RData" and "labels.RData".

Everything seems to be running fine but in the end the "Signature" object is not created (apparently "results/CycISC_lrTest.csv" is not being created...).

Would it be possible to look into this? Thanks a million!

This is the log:

[1] "finished log2"
[1] "CycISC"
[1] "finished cells.symbol.list2"
[1] "finished cells.coord.list2"
[1] "finished cells.symbol.list1"
[1] "finished cells.coord.list1"
[1] "finished data.used.log2.ordered"
[1] "finished group.v"
[1] "finished log2.stat.result"
[1] "finished m.auc"
[1] "finished bigtable"
[1] "finished filter bigtable by log2fc"
[1] "finished filter data.1"
[1] "finished filter data.2"
[1] "finished get log2fold_change"
[1] "finished get counts"
[1] "finished get groups"
No id variables; using all as measure variables
[1] "finished get data_for_MIST"
Assuming data assay in position 1, with name Et is log-transformed.
[1] "finished get FromFlatDF"
[1] "finished get subset"
 Completed [-------------------------------------------------]   1% with 0 failures Completed [>------------------------------------------------]   1% with 0 failures Completed [>------------------------------------------------]   2% with 0 failures
...
Completed [=================================================] 100% with 0 failures                                                                                   
Done!
Fitted zlm on 1388 genes and 3000 cells.
 Using BayesGLMlike ~ Population 
Refitting on reduced model...
 Completed [>------------------------------------------------]   1% with 0 failures Completed [>------------------------------------------------]   2% with 0 failures Completed [>------------------------------------------------]   3% with 0 failures Completed [=>-----------------------------------------------]   3% with 0 failures 
...
Completed [=================================================] 100% with 0 failures                                                                                   
Done!
Error in file(file, ifelse(append, "a", "w")) : 
  cannot open the connection
In addition: Warning messages:
1: 'zlm.SingleCellAssay' is deprecated.
Use 'zlm' instead.
See help("Deprecated") 
2: In file(file, ifelse(append, "a", "w")) :
  cannot open file 'results/CycISC_lrTest.csv': No such file or directory

Kind regards,
Francisco

When following the tutorial, when calling source("../Deconvolution_functions"), it gave the error message:

"Loading required package: MAST
Error: With R version 3.5 or greater, install Bioconductor packages using BiocManager; see https://bioconductor.org/install"

Which was solved by replacing:
"source("https://bioconductor.org/biocLite.R")
biocLite("MAST")"

in the source by:
"BiocManager::install("MAST")"

It might make sense to do it this way in the first place in your package?
Thanks

Thanks for this amazing tool!!

I have one problem using this tool which is when I'm doing buildSignaturesMatrixMAST it ended with this error "could not find function "buildSignaturesMatrixMAST"" even though i have successfully installed MAST package into R. How can I fix it? any help I appreciate in advance.

Bugie
PhD student
at Academia Sinica, Taiwan

Hi
I am using the buildSignatureMatrixMAST() function and I am getting the following error :

This is the traceback :

Error in dimnames(x) <- dn : 
  length of 'dimnames' [2] not equal to array extent 
3.
`colnames<-`(`*tmp*`, value = unique(id)) 
2.
`colnames<-`(`*tmp*`, value = unique(id)) at Deconvolution_functions.R#384
1.
buildSignatureMatrixMAST(scdata = data, id = as.character(phenodata$cellType), 
    path = ".", diff.cutoff = 0.5, pval.cutoff = 0.01) 

The scdata is a gene x cell name matrix .
id is the cell type that each of the cell belongs to .

Please help me fix this error !

Thanks ,
RK

Hi there,

I am having error while using the buildSignatureMatrixMAST function. So it is not able to find the function. Could you help me about it?

Thanks,
Hasan

This error message appears
Error in eval(parse(text = temp)) :
object 'cluster_lrTest.table.BBOX1' not found
It only appears after running the script for a while. Not sure what to do?

Hi,

Thank you for this amazing software. I'm using DWLS to run deconvolution. In your codes, DWLS will perform DE analysis to identify markers for each cell type. Does it mean the input file should be UMI counts, not any nomalized data? Can I use TPM or CPM as the input file?
Thank you.

Best regards

Hi there,
Very useful pipeline and thanks for your great contribution. I am not familiar with the math part in the pipeline, just wondering whether the estimated fractions would be affected by sample size of bulkData. For example, when to input a total of 100 bulk samples and just 50 of them, are the results of those 50 samples from two times are the same or comparable？
Many thanks.

Hi there,

I am currently trying to install the DWLS package, but keep getting the error:

Error: object 'zlm.SingleCellAssay' is not exported by 'namespace:MAST'

Because of the newer versions of MAST no longer containing the zlm.SingleCellAssay function ([https://bioconductor.org/packages/devel/bioc/manuals/MAST/man/MAST.pdf] page 34) and being replaced with zlm.

Is there a way of working around this? Or will it be a case of having to install an older version of MAST (v 0.931 for example). If so, do you know of any repositories which contain the old code?

Thank you!

Sam

Hi

I am using raw count data from single cell RNA-seq and bulk RNA-seq.
I am currently stuck at running the deconvolution function solveDampenedWLS().
I get the error :

Error in solve.QP(D, d, A, bzero) :
  matrix D in quadratic function is not positive definite!

I tried to solve this using nearPD() to get the nearest positive definite matrix for the data , but I still get the same error.

Can you suggest a way I can fix this issue?

Thanks a lot!

I can build my matrix with scRNAseq reference with 12 out of the 28, but I feel getting this error while doing buildSignatureMatrixUsingSeurat at the same point. The connection doesn't actually seem broken as it keeps "breaking" at the same point. I know I can read and write into the file because it has successfully done it 12 file before creating the signature for 12 other cell types.
Any ideas?

Error in file(file, ifelse(append, "a", "w")) :
cannot open the connection
Calls: buildSignatureMatrixMAST ... write.csv -> eval.parent -> eval -> eval -> -> file
In addition: There were 35 warnings (use warnings() to see them)
Execution halted

Hi,

I keep getting the error:
Error in [.data.frame(data.used.log2, , cells.coord.list1) :
undefined columns selected

My sc data matrix is genes(rows) x cell (columns) and my id is cell name (rows) x cell type (column). There are both in data frames. My bulk data is genes (rows) x samples (bulk data samples).

Please help!

Best,
Sara
PhD student at Imperial College London

Dear DWLS team,

Thanks for making the package. I am interested in what bulk RNA-seq normalization method could be used for this package. In the paper, I noticed that you all used FPKM for bulk. You also mentioned that the single-cell normalization is critical. For bulk, have you tested other normalizations, such as raw counts or TPM/CPM?

Thanks,

qingnan

Dear DWLS developer,

We found your method appealing and we would like to use it. The problem is that we don't want to use Seurat DE genes as input but rather an input list we would provide. How would you advice us to proceed ?

Have a nice day,

best,

A. Coudray
PhD student at EPFL
Switzlerland