0
$\begingroup$

I used ComBat() for batch effect correction in my expression data. basically, that function inputs are expression data, Batch covariate, and Model matrix for the outcome of interest and other covariates besides batch. So, I prepared all inputs based on the instruction of ComBat() function in 'sva' package. "Pheno_LMS" is my clinical data inclus batchId of my samples and "TCGA_expr_log" is my expression data. my code is below:

Pheno_LMS structure:

Sample        batchId   age  . . . 
GSM71019.CEL    396     63
GSM71020.CEL    396     58
GSM71021.CEL    410     85
GSM71020.CEL    411     58
GSM71021.CEL    410     40
.
.
.
dim(Pheno_LMS)
[1] 74 37

TCGA_expr_log structure:

                   GSM71019.CEL 1      GSM71019.CEL 2        GSM71019.CEL 3  . . . 
    Gene id          
  ENSG00000000003     10.115170           8.628044                  8.779235  
  ENSG00000000005     5.345168            5.063598                   5.113116
  ENSG00000000009     6.348024            6.663625                   6.465892     
    .
    .
    .
dim(TCGA_expr_log)
[1] 42817    74

batch = Pheno_LMS$batchId
modcombat = model.matrix(~1, data=Pheno_LMS)
TCGA_expr_Co <-  ComBat(as.matrix(TCGA_expr_log),batch = batch,mod = modcombat, par.prior = TRUE,mean.only = TRUE)

but when I run that function, I got below error:

Error in apply(dat[, batch == batch_level], 1, function(x) { : 
  dim(X) must have a positive length

Now, I need anybody comments about my problem. thanks

P.S: when i call batch I see:

> batch
 [1] 13 13 13 14 14 14 14 14 14 14 14 6  6  7  7  7  7  7  7  7  7  7  7  7  7  7  7  7 
[29] 7  8  7  8  11 11 11 14 12 1  6  4  2  3  3  3  3  6  6  1  3  9  8  6  11 7  3  5 
[57] 4  5  5  4  4  11 8  8  12 12 12 12 12 10 10 10 11 13
Levels: 1 2 3 4 5 6 7 8 9 10 11 12 13 14
$\endgroup$
  • 2
    $\begingroup$ dat[, batch == batch_level] is likely empty. $\endgroup$ – Devon Ryan Dec 2 '19 at 8:30
  • $\begingroup$ Thanks for @DevonRyan comment. at the end of my post, I add "batch" content. As you see, when I call batch, for each sample I see a batch value. I have 14 batches.based on your knowledge how can I solve this problem? $\endgroup$ – Mohammad Dec 2 '19 at 15:06
  • $\begingroup$ You need to show us something that’s reproducible. As is we need to read your mind to figure out exactly what’s going wrong. $\endgroup$ – Devon Ryan Dec 2 '19 at 16:37
  • $\begingroup$ Dear @DevonRyan, for finding problem I save "Pheno_LMS" and "TCGA_expr_log" to "my_data.RData" which you can download it from this link: dropbox.com/s/is4b9laou305d5k/my_data.RData?dl=0 $\endgroup$ – Mohammad Dec 2 '19 at 21:35
  • $\begingroup$ Dear @DevonRyan, I appreciate it if you check my_data.RData and guide me for solving the problem. $\endgroup$ – Mohammad Dec 2 '19 at 21:38
1
$\begingroup$

batch = Pheno_LMS$batchId should be batch = Pheno_LMS$BatchId, if you fix that then at least the code you showed works just fine.

| improve this answer | |
$\endgroup$
  • $\begingroup$ Thanks @DevonRyan.In addition of your comment I found 2 batches by just 1 sample. So, I removed them from my TCGA_expr_log and now ComBat() runs perfectly. $\endgroup$ – Mohammad Dec 3 '19 at 18:27
0
$\begingroup$

as suggested in the comment you should verify if the command:

dat[, batch == batch_level]

returns an empty object or not. Apparently, batch and batch_level do not have anyting that match and the function tries to run over an empty dataframe.

| improve this answer | |
$\endgroup$
  • $\begingroup$ Thanks for @gabt comment. in the end of my post, I add "batch" content. As you see, when I call batch, for each sample I see a batch value. i have 14 batches. $\endgroup$ – Mohammad Dec 2 '19 at 15:04

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.