Question: Number of SVA increased after ComBat
gravatar for jonellevillar
18 months ago by
jonellevillar0 wrote:

Dear Friends,

I needed to run two different models with different technical variables as I got an error when I ran them together (Slide and Plate).  


mod1 <- model.matrix(~ AP1 + Age + factor(Gender) + factor(smoker) + factor(Slide), data = targets)

mod0 <- model.matrix(~ Age + factor(Gender) + factor(smoker) + factor(Slide), data = targets) <-, mod1, method = "be")  #SVA = 17

mod2 <- model.matrix(~ AP1 + Age + factor(Gender) + factor(smoker) + factor(Plate) + Array, data = targets)

mod02 <- model.matrix(~ Age + factor(Gender) + factor(smoker) + factor(Plate) + Array, data = targets) <-, mod2, method = "be")  #SVA = 8

I then ran ComBat twice:

modcombat <- model.matrix(~1, data = targets)

adj1 <- ComBat(dat = data, batch = targets$Slide, mod = NULL)

adj2 <- ComBat(dat = adj1, batch = targets$Plate, mod = NULL)

I checked the surrogate variables after adjustment, removing Slide and Plate:

mod3 <- model.matrix(~ AP1 + Age + factor(Gender) + factor(smoker) + Array, data = targets)

mod03 <- model.matrix(Age + factor(Gender) + factor(smoker) + Array, data = targets) <-, mod3, method = "be")  #SVA = 57

I am concerned about the 57 SVA after running ComBat twice.  Are there any ideas or suggestions how I might resolve this problem?  I should mention that my sample size is n = 160.  I am wondering if the problem with increased number of hidden variables comes from the fact that some of the slides only have one case, while other slides have three cases.  One of my colleagues ran the same model with no problem however the sample size was n = 850. 

Thank you for your help.

Jonelle Villar, RN

> sessionInfo()

R version 3.4.4 (2018-03-15)

Platform: x86_64-w64-mingw32/x64 (64-bit)

Running under: Windows >= 8 x64 (build 9200)


Matrix products: default



[1] LC_COLLATE=English_United States.1252  LC_CTYPE=English_United States.1252   

[3] LC_MONETARY=English_United States.1252 LC_NUMERIC=C                          

[5] LC_TIME=English_United States.1252    


attached base packages:

[1] stats    graphics  grDevices utils     datasets methods   base     


other attached packages:

 [1] sva_3.26.0          BiocParallel_1.12.0 genefilter_1.60.0   mgcv_1.8-23         nlme_3.1-131.1     

 [6] forcats_0.3.0       stringr_1.3.0       dplyr_0.7.4         purrr_0.2.4         readr_1.1.1        

[11] tidyr_0.8.0         tibble_1.4.2        ggplot2_2.2.1       tidyverse_1.2.1    


loaded via a namespace (and not attached):

 [1] Rcpp_0.12.16         lubridate_1.7.2      lattice_0.20-35      digest_0.6.15        assertthat_0.2.0    

 [6] psych_1.8.3.3        R6_2.2.2             cellranger_1.1.0     plyr_1.8.4           stats4_3.4.4        

[11] RSQLite_2.1.0        httr_1.3.1           pillar_1.2.1         rlang_0.2.0          lazyeval_0.2.1      

[16] readxl_1.0.0         rstudioapi_0.7       annotate_1.56.2      blob_1.1.1           S4Vectors_0.16.0    

[21] Matrix_1.2-13        splines_3.4.4        foreign_0.8-69       RCurl_1.95-4.10      bit_1.1-12          

[26] munsell_0.4.3        broom_0.4.4          compiler_3.4.4       modelr_0.1.1         pkgconfig_2.0.1     

[31] BiocGenerics_0.24.0  mnormt_1.5-5         matrixStats_0.53.1   IRanges_2.12.0       XML_3.98-1.10       

[36] crayon_1.3.4         bitops_1.0-6         grid_3.4.4           jsonlite_1.5         xtable_1.8-2        

[41] gtable_0.2.0         DBI_0.8              magrittr_1.5         scales_0.5.0         cli_1.0.0           

[46] stringi_1.1.7        reshape2_1.4.3       bindrcpp_0.2.2       limma_3.34.9         xml2_1.2.0          

[51] tools_3.4.4          bit64_0.9-7          Biobase_2.38.0       glue_1.2.0           hms_0.4.2           

[56] survival_2.41-3      parallel_3.4.4       yaml_2.1.18          AnnotationDbi_1.40.0 colorspace_1.3-2    

[61] rvest_0.3.2          memoise_1.1.0        bindr_0.1.1          haven_1.1.1  



combat sva • 202 views
ADD COMMENTlink modified 18 months ago • written 18 months ago by jonellevillar0
Please log in to add an answer.


Use of this site constitutes acceptance of our User Agreement and Privacy Policy.
Powered by Biostar version 16.09
Traffic: 290 users visited in the last hour