I am using the voomLmFit workflow to process and analyze an RNA-seq dataset. This goes fine.

Yet, when I try to subset the MArrayLM object, that is obtained by running voomLmFit, to only include a subset of genes (in my case annotated with a specific biotype) I noticed that the content of the EList component is not subsetted (= $E, $weights and $genes). This is in contrast to what I expected.

Therefore: is this intentional? If so, how then to subset the EList component as well?

Thanks,

G

Note 1: I observed the same when subsetting the fit object that is obtained after the eBayes() step.

Note 2: is there a reason why the annotation information is present twice? $genes is also present in the EList component.

Below some code to illustrate this (as per Chapter 4 (page 60) of the edgeR user guide).

> library(edgeR)
> library(tweeDEseqCountData)
> data(pickrell1)
> 
> Counts <- exprs(pickrell1.eset)
> Gender <- pickrell1.eset$gende
> 
> data(annotEnsembl63)
> annot <- annotEnsembl63[,c("Symbol","Chr")]
> 
> y <- DGEList(counts=Counts, genes=annot[rownames(Counts),])
> 
> isexpr <- filterByExpr(y, group=Gender)
> hasannot <- rowSums(is.na(y$genes))==0
> 
> y.filtered <- y[isexpr & hasannot, , keep.lib.sizes=FALSE]
> 
> y.filtered <- normLibSizes(y.filtered)
> 
> design <- model.matrix(~Gender)
> 
> v <- edgeR::voomLmFit(counts = y.filtered, design = design, block = NULL, sample.weights = FALSE, plot = TRUE, normalize.method = "none")
> 
> ## define some genes of interest
> my.genes <- c("ENSG00000127720", "ENSG00000051596", "ENSG00000236211")
> 
> ## subset.
> ## Note that all content of the EList still has original dimension 
> v[my.genes,]
An object of class "MArrayLM"
$coefficients
                (Intercept)  Gendermale
ENSG00000127720   0.6867804 -0.09056663
ENSG00000051596   5.1102007  0.01717677
ENSG00000236211   5.2132901  0.03700956

$stdev.unscaled
                (Intercept) Gendermale
ENSG00000127720  0.12254247  0.1899450
ENSG00000051596  0.06683824  0.1030391
ENSG00000236211  0.06678426  0.1029758

$sigma
[1] 0.7928014 1.0189892 2.1298141

$df.residual
[1] 67 67 67

$cov.coefficients
            (Intercept)  Gendermale
(Intercept)       0.025 -0.02500000
Gendermale       -0.025  0.05948276

$pivot
[1] 1 2

$rank
[1] 2

$Amean
ENSG00000127720 ENSG00000051596 ENSG00000236211 
      0.6447488       5.1157701       5.2302451 

$method
[1] "ls"

$design
  (Intercept) Gendermale
1           1          1
2           1          1
3           1          0
4           1          1
5           1          0
64 more rows ...

$genes
                    Symbol Chr
ENSG00000127720   C12orf26  12
ENSG00000051596      THOC3   5
ENSG00000236211 AC068137.5   2

$targets
        group lib.size norm.factors
NA18486     1  7750614    0.9294587
NA18498     1 13614927    1.0964857
NA18499     1  8570996    0.9576338
NA18501     1  8596932    1.1944060
NA18502     1 13377004    0.9422220
64 more rows ...

$EList
$E
                   NA18486  NA18498  NA18499  NA18501     NA18502   NA18504
ENSG00000127720 -0.1483337 1.122363 0.820972 1.789632  0.20215888 0.1124152
ENSG00000051596  5.8254189 4.993297 4.747626 4.842008  4.69069161 4.5306694
ENSG00000236211  6.4629749 5.313100 5.344534 4.591169  5.07719320 5.5950791
ENSG00000213697  1.5087786 1.542939 1.092274 2.049276 -0.01196593 0.9295512
ENSG00000135541  0.7950828 1.945486 1.690911 1.662252  1.91403349 1.9099224
                   NA18505   NA18507   NA18508   NA18510   NA18511   NA18516
ENSG00000127720 0.89663383 0.4033372 0.3556554 0.2756946 1.0557263 1.4455556
ENSG00000051596 5.15742827 5.5926549 5.5047867 5.8904044 4.7865752 5.1568699
ENSG00000236211 4.55879084 4.4194312 5.2015256 4.8111551 4.7697604 4.7508324
ENSG00000213697 1.45056943 1.0486723 1.1939044 1.1687793 0.8147182 0.2697057
ENSG00000135541 0.09693248 1.5273259 0.9836866 1.8606571 0.1626415 0.8723702
                   NA18517   NA18519   NA18520  NA18522    NA18523   NA18852
ENSG00000127720  0.4625639 0.3238008 0.3399790 1.307367 0.02352065 0.6362799
ENSG00000051596  5.7459651 5.3511718 5.5723555 5.943635 5.28176545 4.7924555
ENSG00000236211  5.8669000 5.7238539 5.8693372 4.045515 5.49186699 5.0015355
ENSG00000213697 -0.3545721 1.4356937 0.6294856 0.234611 0.95640645 0.8333168
ENSG00000135541  2.0986005 1.4356937 1.7749161 1.680411 0.95640645 1.7544614
                 NA18853   NA18855   NA18856   NA18858  NA18861   NA18862
ENSG00000127720 0.417341 0.3112431 1.2378376 0.9499893 1.584091 1.7294831
ENSG00000051596 4.482494 5.8950029 4.9620749 4.6024147 5.095053 4.7791139
ENSG00000236211 7.341379 6.1278428 4.7113788 5.0410589 4.454182 5.6892533
ENSG00000213697 1.023062 1.0249389 0.2763117 0.7529524 1.518503 0.3586454
ENSG00000135541 1.385632 1.6446669 1.0727783 0.8216652 2.321057 2.6805735
                 NA18870   NA18871   NA18909    NA18912    NA18913   NA18916
ENSG00000127720 1.327554 0.9402908 0.6780392  1.5197835 -0.4385061 1.3457396
ENSG00000051596 4.627834 4.2100962 4.8572540  4.3921357  5.2894143 4.5584936
ENSG00000236211 4.537837 4.8533733 4.6632855  6.1144091  5.8823608 5.1618012
ENSG00000213697 1.327554 1.0304886 0.1097555 -0.6270579  1.1464564 0.6725499
ENSG00000135541 1.554325 1.8687376 1.3178637  1.5197835  0.6834844 1.8029466
                   NA19093    NA19098   NA19099   NA19101   NA19102  NA19108
ENSG00000127720  1.1236172 0.09435953 0.5798838 1.3069284 0.3527364 1.185508
ENSG00000051596  4.9649194 5.55472766 5.1873785 4.5516440 4.6912157 5.220891
ENSG00000236211  3.2683887 4.51004095 3.4144601 5.0229145 5.7317815 4.888985
ENSG00000213697 -0.4613453 1.50423533 1.2778552 0.3666346 0.7397595 1.018051
ENSG00000135541  3.1316099 1.81056657 2.0538149 2.0996958 1.8617500 2.029106
                  NA19114   NA19116   NA19119   NA19127    NA19128  NA19130
ENSG00000127720 0.2312883 0.7382099 0.4768217 0.5111366 0.50770071 1.302568
ENSG00000051596 4.8860948 4.8857938 4.6892713 5.7533002 5.25032293 5.249536
ENSG00000236211 7.6223785 5.9443410 5.4523390 6.2257984 6.53386843 5.073173
ENSG00000213697 0.5938584 0.3930744 2.4488073 0.5111366 0.02227388 1.261345
ENSG00000135541 0.4828271 2.6993209 2.0617842 2.7976593 2.34420198 2.465703
                   NA19131   NA19137   NA19138   NA19140   NA19143    NA19144
ENSG00000127720  0.5344892 0.6828116 0.3746129 0.1156107 0.3226062 -0.8999756
ENSG00000051596  5.0864446 5.7960497 4.8981749 5.4574383 4.9744002  4.9939093
ENSG00000236211  5.5511383 4.7702745 4.2697980 6.0108774 4.5757609  5.8205975
ENSG00000213697 -0.5073310 0.5865963 1.5200434 0.5701765 0.5090193  1.4219525
ENSG00000135541  2.4386292 2.4967997 1.7052582 1.4692476 1.3953625  1.5081091
                    NA19147   NA19152   NA19153   NA19159    NA19160  NA19171
ENSG00000127720  0.10740287 1.0880426 0.1875934 -1.146275 0.04797117 1.187973
ENSG00000051596  5.15559355 5.3903732 5.0861687  4.944216 4.79889522 4.872844
ENSG00000236211  5.11327146 4.5885209 5.0828291  5.511937 4.67462662 3.995328
ENSG00000213697 -0.02384167 0.8921224 1.3068923  1.718796 0.96550901 1.489628
ENSG00000135541  1.77366547 1.7689741 2.4485535  1.175653 1.70247460 2.136747
                   NA19172  NA19190    NA19192    NA19193   NA19200   NA19201
ENSG00000127720  0.4772295 0.340551 -0.1694657  0.3290738 0.6201132 0.7311505
ENSG00000051596  4.8541989 4.606588  5.1678377  5.6866258 4.8792462 4.4067156
ENSG00000236211  5.0284037 4.365580  6.2527994  8.3278482 4.4312798 5.4303567
ENSG00000213697 -0.5772183 0.340551  2.4494442 -1.0245632 2.8539104 2.3373295
ENSG00000135541  1.1486067 1.122960  2.0381298  1.4779372 2.0550503 2.2050817
                   NA19203   NA19204     NA19209    NA19210   NA19222   NA19225
ENSG00000127720  0.0324493 1.3271402 -0.16534375 -0.1937652 1.4099509 1.3273477
ENSG00000051596  5.4905688 4.9697681  5.25284620  6.0165356 5.7410975 5.4573800
ENSG00000236211  5.5446762 4.5240954  5.33418327  6.9161222 4.4688446 4.9101226
ENSG00000213697 -0.2190895 0.5274389 -0.05431244  1.0640326 0.8794361 0.8313902
ENSG00000135541  2.9959234 2.0393379  1.93226705  2.3160959 2.4995881 1.8497687
                  NA19238   NA19239    NA19257
ENSG00000127720 1.4395553 1.8085983  1.1312497
ENSG00000051596 5.2685187 5.3014033  5.3920442
ENSG00000236211 4.9710466 4.5013260  4.7540804
ENSG00000213697 0.7816613 0.8228121 -0.1754116
ENSG00000135541 2.5471960 2.4991150  1.8434474
17512 more rows ...

$weights
         [,1]     [,2]     [,3]     [,4]     [,5]     [,6]     [,7]     [,8]
[1,] 1.251586 1.942032 1.399736 1.541965 1.815790 1.491340 1.817455 1.926966
[2,] 5.489630 5.624351 5.563001 5.647866 5.653545 5.635361 5.653470 5.627023
[3,] 5.545679 5.603760 5.598851 5.655623 5.644411 5.651818 5.644176 5.607108
[4,] 1.570496 2.456153 1.340979 1.954851 1.735290 1.887949 1.736874 2.438297
[5,] 2.120196 3.224496 2.173426 2.617118 2.792673 2.534869 2.795030 3.202901
         [,9]    [,10]    [,11]    [,12]    [,13]    [,14]    [,15]    [,16]
[1,] 1.841693 1.394420 2.077705 1.625174 1.888593 1.728688 1.638039 1.723509
[2,] 5.652383 5.596100 5.617030 5.655631 5.646163 5.653827 5.651471 5.654075
[3,] 5.640776 5.629390 5.597939 5.654544 5.633837 5.643307 5.656543 5.644083
[4,] 1.759935 1.759572 1.984606 2.064171 1.804553 2.197417 1.566876 2.190728
[5,] 2.828530 2.373408 3.148111 2.749081 2.893043 2.909171 2.539123 2.901225
        [,17]    [,18]    [,19]    [,20]    [,21]    [,22]    [,23]    [,24]
[1,] 1.101696 1.867750 1.388510 1.582176 1.314756 1.776112 1.328133 1.634925
[2,] 5.222958 5.648995 5.593202 5.640857 5.544903 5.655353 5.509792 5.656303
[3,] 5.300989 5.637186 5.627773 5.652638 5.590500 5.650074 5.554565 5.654044
[4,] 1.057593 1.784725 1.751739 1.513935 1.654016 1.697770 1.272995 2.076723
[5,] 1.676916 2.864399 2.363470 2.456826 2.234534 2.736814 2.055733 2.764462
        [,25]    [,26]    [,27]    [,28]    [,29]    [,30]    [,31]    [,32]
[1,] 1.513844 1.337407 1.889688 1.101054 1.653474 1.725318 1.606082 2.059310
[2,] 5.622091 5.560858 5.646015 5.221811 5.657514 5.657258 5.646413 5.601891
[3,] 5.641065 5.602288 5.633637 5.299991 5.653104 5.653968 5.654328 5.576847
[4,] 1.449172 1.684024 1.805594 1.056983 2.100566 1.649570 1.536572 2.592129
[5,] 2.353998 2.274371 2.894547 1.675847 2.793695 2.665115 2.492654 3.388496
        [,33]    [,34]    [,35]    [,36]    [,37]    [,38]     [,39]    [,40]
[1,] 2.028913 1.894343 1.064646 2.111553 1.472951 1.946478 0.9809772 1.722634
[2,] 5.625165 5.632875 5.155325 5.610860 5.603624 5.638489 5.0632560 5.657085
[3,] 5.608138 5.614278 5.237313 5.590924 5.630788 5.623479 5.1697116 5.654097
[4,] 1.938229 2.399795 1.022230 2.017374 1.410426 1.859430 1.2171923 1.647023
[5,] 3.082920 3.156285 1.615403 3.192700 2.290006 2.972348 1.6199705 2.661320
        [,41]    [,42]    [,43]    [,44]    [,45]    [,46]    [,47]    [,48]
[1,] 1.722194 1.897934 1.768640 1.542089 1.758088 1.802856 1.724192 1.706720
[2,] 5.654138 5.632224 5.655699 5.631323 5.652437 5.654129 5.657186 5.654885
[3,] 5.644281 5.613627 5.651159 5.647984 5.638959 5.646240 5.654022 5.646620
[4,] 2.189030 2.404035 1.690681 1.475954 2.233904 1.723065 1.648501 2.169050
[5,] 2.899207 3.161422 2.726280 2.396524 2.954225 2.774476 2.663523 2.875451
        [,49]    [,50]    [,51]    [,52]    [,53]    [,54]    [,55]    [,56]
[1,] 1.638906 2.008920 1.758545 1.353010 1.786382 1.465868 1.546138 1.562827
[2,] 5.651530 5.628552 5.652416 5.531585 5.648654 5.628888 5.632299 5.636291
[3,] 5.656602 5.612386 5.638892 5.571910 5.634516 5.648347 5.648963 5.651248
[4,] 1.567698 1.919141 2.234469 1.296624 2.268826 1.854255 1.479792 1.495613
[5,] 2.540383 3.056165 2.954926 2.096715 2.996569 2.493475 2.402617 2.427727
        [,57]    [,58]    [,59]    [,60]    [,61]    [,62]    [,63]    [,64]
[1,] 1.678833 1.493514 1.404808 1.899952 1.714734 1.514104 1.940458 1.794416
[2,] 5.656249 5.612976 5.601162 5.644636 5.654497 5.622207 5.639281 5.647503
[3,] 5.650892 5.635995 5.632213 5.631776 5.645405 5.641130 5.624548 5.632962
[4,] 2.133169 1.429904 1.773346 1.815358 2.179397 1.449419 1.853673 2.278738
[5,] 2.832716 2.322432 2.390877 2.908637 2.887757 2.354390 2.964226 3.008489
        [,65]    [,66]    [,67]    [,68]    [,69]
[1,] 1.310924 1.665428 1.858824 1.899941 1.643896
[2,] 5.494107 5.653325 5.650220 5.631861 5.651870
[3,] 5.542387 5.656895 5.638408 5.613172 5.656943
[4,] 1.256673 1.592842 1.776233 2.406406 1.572429
[5,] 2.026921 2.578892 2.852119 3.164294 2.547632
17512 more rows ...

$genes
                      Symbol Chr
ENSG00000127720     C12orf26  12
ENSG00000051596        THOC3   5
ENSG00000236211   AC068137.5   2
ENSG00000213697 RP11-513I7.2  11
ENSG00000135541         AHI1   6
17512 more rows ...


> 
> 
> packageVersion("edgeR")
[1] '4.2.1'
> 

Yes, the current behaviour is absolutely intentional. It doesn't seem to me to be meaningful or useful to try to subset the EList object automatically with the MArrayLM object.

The purpose of voomLmFit() is to mimic the classic limma-voom pipeline that might go:

v <- voom(y, design)
fit <- lmFit(v, design)

Here, v is an EList object and fit is an MArrayLM object. You can subject both v and fit in the usual limma way. They have the same rows but different columns so they are not conformal objects. You would obviously subset them independently.

If v contains gene annotation in v$genes, then the same annotation is carried through to fit$genes.

It is the same with voomLmFit(). Now you have

fit <- voomLmFit(y, design)

which is similar to the classic pipeline except that now v is contained in fit$EList. fit$EList is not intended to be part of the MArrayLM object but rather is a way to store the intermediate logCPM values and weights in case you need them for another purpose. fit obeys all the usual rules of an MArrayLM object and fit$EList obeys all the usual rules of an EList object. It is best to think of them as separate objects, as in the classic pipeline, rather than as some sort of new composite class of object.

If you did want to subset fit$Elist, then you can use fit$EList[i,j], which would be just the same as v[i,j].

If you won't want the EList, then use

fit <- voomLmFit(v, design, keep.EList=FALSE)

and then fit will only contain the MArrayLM components.

You can just create a wrapper to do that if you so choose.

> fun <- function(malst, genes) {
 malst$EList <- malst$EList[genes,]
 malst[genes,]
 }

> newv <- fun(v, my.genes)

It's my understanding that the EList object is included in case you want the normalized logCPM values and weights for an additional purpose. It's not used by any of the functions meant to operate on an MArrayLM object, so I am not surprised that [.MArrayLM doesn't also subset the EList object.