R Deseq2 Error: Inv(): Matrix Appears To Be Singular

Question

[komal.rathi]

Hi everyone,

I am trying to use DESeq2 to perform differential gene expression on my HTSeq count data.

Code:

 > dim(countData) #my count data
 [1] 13875    64

 > head(countData)
                   C00060 C00079 C00135 C00150 C00154
ENSG00000005206.12    721    759    966    375    285
ENSG00000006062.9     542    442    735    238    552
ENSG00000012171.13    635    352    813    316    391
ENSG00000031544.10      7      3     14      2      0
ENSG00000034063.9      21      8     34     77     68
ENSG00000049319.2       0      1      3      1      0

 #create an object of type DESeqDataSet, make design taking into consideration the age, status, gender and site.
 > dds <- DESeqDataSetFromMatrix(countData = countData,
                          colData = colData,
                          design = ~0+status+age+gender+site) 

 > colData(dds)$status <- factor(colData(dds)$status,
                             levels=c("NF","ISCH","DCM"))

 > summary(colData) #this is how colData looks. There are 64 samples, 3 different disease states : NF, DCM & ISCH.
 sample_name  status    age             gender      site   
 C00060 : 1   DCM :22   Min.   :21.00   Female:32   Cleveland:24  
 C00079 : 1   ISCH:20   1st Qu.:48.00   Male  :32   Penn     :40  
 C00135 : 1   NF  :22   Median :56.00                             
 C00150 : 1             Mean   :51.97                             
 C00154 : 1             3rd Qu.:58.00                             
 C00176 : 1             Max.   :67.00                             
 (Other):58 

 # running DESeq() gives an error
 > dds <- DESeq(dds,betaPrior=F)
 estimating size factors
 estimating dispersions
 gene-wise dispersion estimates

 Error: inv(): matrix appears to be singular

This is my R sessioninfo()

R version 3.0.1 (2013-05-16)
Platform: x86_64-pc-linux-gnu (64-bit)

locale:
[1] LC_CTYPE=en_US.UTF-8       LC_NUMERIC=C               LC_TIME=en_US.UTF-8        LC_COLLATE=en_US.UTF-8     LC_MONETARY=en_US.UTF-8   
[6] LC_MESSAGES=en_US.UTF-8    LC_PAPER=C                 LC_NAME=C                  LC_ADDRESS=C               LC_TELEPHONE=C            
[11] LC_MEASUREMENT=en_US.UTF-8 LC_IDENTIFICATION=C       

attached base packages:
[1] parallel  stats     graphics  grDevices utils     datasets  methods   base     

other attached packages:
[1] limma_3.18.13           DESeq2_1.2.10           RcppArmadillo_0.4.200.0 Rcpp_0.11.1             GenomicRanges_1.14.4    XVector_0.2.0          
[7] IRanges_1.20.7          BiocGenerics_0.8.0      BiocInstaller_1.12.0   

loaded via a namespace (and not attached):
[1] annotate_1.40.1      AnnotationDbi_1.24.0 Biobase_2.22.0       DBI_0.2-7            genefilter_1.44.0    grid_3.0.1           lattice_0.20-15     
[8] locfit_1.5-9.1       RColorBrewer_1.0-5   RSQLite_0.11.4       splines_3.0.1        stats4_3.0.1         survival_2.37-7      tools_3.0.1         
[15] XML_3.98-1.1         xtable_1.7-3

I am using DESeq2 version 1.2.10 and I found this thread which says upgrade DESeq2 version to 1.3.48 I don't know if it is something to do with the package upgrade or my code. But if it is something to do with the package upgrade, I couldn't find DESeq2 version 1.3.48 so I am unable to upgrade DESeq2. Any help would be appreciated.

Thanks,

[Devon Ryan]

This sounds almost identical to the issue that you linked to that Michael committed a fix toward. One other thing to try would be to not remove the intercept (i.e., design = ~status+age+gender+site). Given the paramaterization you've chosen, you'll want to keep the intercept anyway.

[komal.rathi]

Yes I tried both ways keeping it and without it, gives me the error anyway. How do I upgrade DESeq2 when there is no update available for it.

[Devon Ryan]

The devel branch (Bioconductor 2.14) is currently at version 1.3.76, which contains the fix, so you could switch over to that if you really want. However, I think the next version of Bioconductor is scheduled to be released by Monday (the 14th) anyway, so you might do yourself a favour and just wait until then to upgrade.

[komal.rathi]

Thank you for the suggestion! I will re-run my analysis on Monday and if everything goes smoothly, I will post it as an answer below.

[Michael Love]

Let me know if you continue to see this with v1.4. thanks.