Reza • 10

@Reza-24551

Last seen 4 days ago

I run the same commands on the same data in Rstudio in Windows and Linux, but I get different results in step removing batch effect using “Combat” function (sva package). Running “Combat” on Windows is error-free, while on Linux I get the following error:

Error in ((dat - t(design %*% B.hat))^2) %*% rep(1/n.array, n.array) : 
requires numeric/complex matrix/vector arguments

Commands

    all <- read.delim("All.txt", header = T, row.names = 1) 
    allm <- all - rowMeans(all)
    pc <- prcomp(allm)
    Sets <- (c(rep("GSEx", 12), rep("GSExx", 12), rep("GSExxx", 9), rep("GSExxxx", 13)))
    pcr <- data.frame(pc$r[,1:3], Sets)
    allc <- ComBat(all, Sets)

For reproducibility

> dput(head(all))
structure(list(A1 = c(6.9381, 7.5079, 7.2303, 8.9908, 8.2564, 
11.0564), A2 = c(6.9594, 7.3883, 7.1023, 8.9328, 8.1679, 10.8513
), A3 = c(6.8419, 7.2109, 6.9548, 8.7414, 8.0299, 10.8115), A4 = c(6.6619, 
7.1364, 6.887, 8.6863, 8.006, 10.8775), A5 = c(6.9296, 7.2845, 
7.0382, 8.8176, 8.1234, 10.9084), A6 = c(7.1549, 7.5539, 7.2842, 
9.0999, 8.3527, 11.1642), A7 = c(6.6192, 7.1043, 6.9482, 8.6802, 
7.9977, 10.7216), A8 = c(7.2778, 7.6654, 7.3542, 9.2437, 8.5063, 
11.2892), A9 = c(7.282, 7.5827, 7.3097, 9.116, 8.4326, 11.1977
), A10 = c(7.3402, 7.7757, 7.4926, 9.2278, 8.6172, 11.1968), 
    A11 = c(6.6635, 7.1918, 6.8924, 8.6318, 7.9105, 10.7312), 
    A12 = c(6.8926, 7.3513, 6.9834, 8.7605, 8.0945, 10.8734), 
    B1 = c(6.8014, 7.1908, 6.8015, 8.8403, 8.0259, 10.7076), 
    B2 = c(6.5855, 7.1594, 6.6593, 8.7153, 7.9119, 10.7066), 
    B3 = c(6.4392, 7.0688, 6.5803, 8.6158, 7.8153, 10.6042), 
    B4 = c(6.7046, 7.078, 6.7047, 8.6752, 7.8961, 10.6192), B5 = c(7.9128, 
    8.3322, 8.0958, 9.8519, 8.9967, 11.6034), B6 = c(6.9226, 
    7.4705, 7.1007, 8.9497, 8.2543, 10.711), B7 = c(6.421, 7.0791, 
    6.4662, 8.5758, 7.7724, 10.5219), B8 = c(6.4972, 7.1769, 
    6.7016, 8.7424, 7.9157, 10.7647), B9 = c(6.6425, 7.3366, 
    6.5679, 8.8394, 7.9696, 10.7032), B10 = c(6.3134, 6.7753, 
    6.4695, 8.4463, 7.644, 10.5063), B11 = c(6.4938, 7.0694, 
    6.336, 8.6302, 7.688, 10.5626), B12 = c(6.2451, 6.9604, 6.2342, 
    8.4628, 7.6297, 10.5476), C1 = c(4.9469, 5.1746, 5.1522, 
    9.5331, 8.9385, 11.4915), C2 = c(5.0953, 5.1003, 5.2505, 
    9.6333, 9.0602, 11.4867), C3 = c(5.0583, 5.0543, 5.1091, 
    9.6526, 9.0479, 11.5055), C4 = c(4.9402, 5.1395, 5.2561, 
    9.655, 9.0221, 11.5286), C5 = c(5.0406, 5.1575, 5.2761, 9.6413, 
    9.0673, 11.5767), C6 = c(4.9726, 5.3945, 5.1122, 9.5541, 
    9.0265, 11.5038), C7 = c(5.3879, 5.1028, 5.0608, 9.803, 9.2769, 
    11.8283), C8 = c(5.127, 5.142, 5.1496, 9.9158, 9.2543, 11.8024
    ), C9 = c(5.3067, 5.1215, 5.0303, 9.9562, 9.3744, 11.8254
    ), D1 = c(7.0772, 7.4086, 7.2877, 9.0927, 8.6348, 11.2268
    ), D2 = c(6.6349, 7.0915, 6.6813, 8.5535, 8.1593, 10.9012
    ), D3 = c(7.4336, 7.7217, 7.512, 9.3543, 8.8803, 11.5023), 
    D4 = c(7.4971, 7.8568, 7.672, 9.441, 9.0128, 11.6245), D5 = c(7.2505, 
    7.5549, 7.3561, 9.2116, 8.6796, 11.3779), D6 = c(6.8906, 
    7.2358, 7.054, 8.8379, 8.4158, 11.0953), D7 = c(6.6697, 7.0378, 
    6.9185, 8.7005, 8.1895, 11.0119), D8 = c(6.955, 7.2885, 7.1322, 
    8.922, 8.4309, 11.1848), D9 = c(7.4286, 7.628, 7.5056, 9.2589, 
    8.8542, 11.5108), D10 = c(7.1027, 7.3881, 7.2015, 9.0158, 
    8.5069, 11.2525), D11 = c(6.9174, 7.2366, 7.1286, 8.8789, 
    8.5092, 11.1913), D12 = c(7.3573, 7.6144, 7.4158, 9.231, 
    8.7868, 11.4215), D13 = c(7.3078, 7.5243, 7.4756, 9.2333, 
    8.7521, 11.4136)), row.names = c("AFFX-BioB-3_at", "AFFX-BioB-5_at", 
"AFFX-BioB-M_at", "AFFX-BioC-3_at", "AFFX-BioC-5_at", "AFFX-BioDn-3_at"
), class = "data.frame")

how can solve this problem?

Thanks for your helps

Kevin Blighe 2.3k

@u16406

Last seen 1 hour ago

Republic of Ireland

Thanks for posting the reproducible data. Using this data, the command runs okay here on Linux:

allc <- sva::ComBat(all, Sets)

Found4batches
Adjusting for0covariate(s) or covariate level(s)
Standardizing Data across genes
Fitting L/S model and finding priors
Finding parametric adjustments
Adjusting the Data

Perhaps you can show the output of your sessionInfo()? - here is mine:

sessionInfo()
R version 4.0.3 (2020-10-10)
Platform: x86_64-pc-linux-gnu (64-bit)
Running under: Ubuntu 16.04.7 LTS

Matrix products: default
BLAS:   /usr/lib/atlas-base/atlas/libblas.so.3.0
LAPACK: /usr/lib/atlas-base/atlas/liblapack.so.3.0

locale:
 [1] LC_CTYPE=pt_BR.UTF-8       LC_NUMERIC=C              
 [3] LC_TIME=en_GB.UTF-8        LC_COLLATE=pt_BR.UTF-8    
 [5] LC_MONETARY=en_GB.UTF-8    LC_MESSAGES=pt_BR.UTF-8   
 [7] LC_PAPER=en_GB.UTF-8       LC_NAME=C                 
 [9] LC_ADDRESS=C               LC_TELEPHONE=C            
[11] LC_MEASUREMENT=en_GB.UTF-8 LC_IDENTIFICATION=C       

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

loaded via a namespace (and not attached):
 [1] Rcpp_1.0.4.6         AnnotationDbi_1.50.0 edgeR_3.30.3        
 [4] BiocGenerics_0.34.0  splines_4.0.3        IRanges_2.22.2      
 [7] bit_1.1-15.2         BiocParallel_1.22.0  lattice_0.20-41     
[10] xtable_1.8-4         rlang_0.4.6          blob_1.2.1          
[13] grid_4.0.3           parallel_4.0.3       nlme_3.1-149        
[16] Biobase_2.48.0       mgcv_1.8-33          DBI_1.1.0           
[19] genefilter_1.70.0    matrixStats_0.56.0   survival_3.2-3      
[22] bit64_0.9-7          digest_0.6.25        Matrix_1.2-18       
[25] sva_3.36.0           vctrs_0.3.1          S4Vectors_0.26.1    
[28] bitops_1.0-6         RCurl_1.98-1.2       memoise_1.1.0       
[31] RSQLite_2.2.0        limma_3.44.3         compiler_4.0.3      
[34] locfit_1.5-9.4       stats4_4.0.3         XML_3.99-0.3        
[37] annotate_1.66.0