Error in getb0.rnaseq()

[RK1912]

Hi
I am running deconvSeq for deconvolving 200 bulk samples.
My current data looks like this :
single cell data : ~10,000 genes and ~2500 unique cell IDs and their corresponding cell types in the metadata.
Also the single cell data is multi subject and the sample names are given under the subjectname column in the metadata.
bulk data : ~10000 genes and 200 individuals.

I get the following error:

Error in dimnames(x) <- dn : 
  length of 'dimnames' [2] not equal to array extent 
3.`colnames<-`(`*tmp*`, value = colnb0) 
2.deconvSeq::getb0.rnaseq(dge.singlecell, design.singlecell, ncpm.min = 1, 
    nsamp.min = 4) at helper_functions.R#668

Please let me know how I can fix this.

Thanks,
RK

[rosedu1]

Can you send me a small example of your data?
Rose

[RK1912]

Hi Rose,
Here is my data:
https://github.com/RK1912/Deconv_data

Thanks ,
RK

[rosedu1]

I am unable to open the files, they are corrupted somehow. You can just send me a toy example with a few lines so that I can check on the formatting.

[RK1912]

hi

Let me know if you have access to the files .
I also provided the script I used for running deconvseq.
https://drive.google.com/drive/folders/1e3JlxFqqyRZKiLHV1wtjJxHAIe5Pthso?usp=sharing

Thanks !

[RK1912]

Hi
Were you able to access the data and run the deconvolution with it ?
Let me know so I can make any corrections if necessary.
RK

[rosedu1]

I am still unable to open them:

load("/Users/rdu/Downloads/meta.rds")
Error in load("/Users/rdu/Downloads/meta.rds") :
bad restore file magic number (file may be corrupted) -- no data loaded
In addition: Warning messages:
1: In readChar(con, 5L, useBytes = TRUE) :
truncating string with embedded nuls
2: file ‘meta.rds’ has magic number 'X'
Use of save versions prior to 2 is deprecated

If you can copy and paste a small portion of your file (e.g. a few lines, a few methylation sites), I can at least see the formatting that you used.
Rose

[vjcitn]

Hi @rosedu1 -- you need to use readRDS not load for .rds files. I am checking into this report now and will get back shortly.

[rosedu1]

I am able to load it now. Will take a look

[vjcitn]

The 'sc' is not counts.

> sc["RLIM", 5:10]
     S027_L6Q3_H01 S027_L5Q2_K14 S010_L1Q2_E08 S033_L6Q4_N18 S006_L2Q4_F08
RLIM       0.47088             0     0.5283183     0.4271253             0
     S006_L1Q1_E21
RLIM     0.4536176

[vjcitn]

The documentation and the example in the package are explicit that the expression data need
to be in the form of counts. However, the error that was produced shows that the software needs
some improvement in a) recognizing when non-count data are supplied, and failing at that point, b)
3.colnames<-(tmp, value = colnb0) error arose when b0 could not be estimated for any
gene. the b0 matrix was completely NA and colnb0 was just a scalar zero. This needs to be
bulletproofed.

[rosedu1]

Thanks. I will add in the error messages