[BioC] HTqPCR: weird error and wrong feature positions after limma.CtData

[Heidi Dvinge]

Hello Andreia,

> Dear all,
>
> I am using HTqPCR to analyse different expression of miRNAs, where I am
> comparing two treatments, 1 with 2 biological replicates and the second
> with
> 2 biological replicates
>
Okay, so you have two samples with two replicates each.

> I have followed the vignete, read the files, made raw.cat and filtered out
> IQR<1.5 and category= undetermined and type=endogenous
>  and everything went well until the fit test:
>
> qDE.limma<-limmaCtData(iqr.filt, design=design, contrasts=contrasts,
> ndups=2, spacing=1)
> Error in dim(M) <- c(spacing, ndups, ngroups, nslides) :
>   dims [product 396] do not match the length of object [402]
>
What's the dimensions of your data, i.e. number of genes on each card?
show(iqr.filt) What does traceback() say?

Using ndups=2 and spacing=1 means that there's a replicate of each gene,
and they're located right next to each other. Is this the case you your
array?

> can someone tell me what this means? it still performed the test but I
> would
> like to clear out this.
>
> The other weird thing is the result of the test:
>
>    genes feature.pos     t.test      p.value  adj.p.value        ddCt
> meanTarget meanCalibrator categoryTarget
> 39   miR-122a       H7;H8  6.7414852 6.141932e-06 0.0002456773  14.8264063
> 34.833906      20.007500   Undetermined
> 1     miR-18a       A1;A2  5.5408137 5.366491e-05 0.0010732983  13.5365104
> 30.431094      16.894583   Undetermined
> 10    miR-368     B11;B12 -4.7841570 2.320152e-04 0.0030935366 -23.1045833
> 7.119063      30.223646   Undetermined
>
> the feature.pos is different in both treatments but it is not, I have
> checked and re-checked the files and miR-18a is in pos. A1 in all cards.
>
This is because of ndups=2, spacing=1 that you've specified when you did
the test. It then says that miR-18a is in position A1, but since it should
be replicated and the replicates are next to each other, it must also be
in A2. Ditto for H7;H8 and B11:B12.

If your spots on the plate are not replicated, you need to set ndups=1. If
they're replicated but not next to each other, you either need to adjust
the spacing parameter accordingly, or sort your file so they are.
?limmaCtdata gives an example of how to reorder data to get the genes in
consecutive rows.

When you filter your data, have you then got the same number of replicates
of each spot afterward? What does table(featureNames(iqr.filt)) say?

HTH
\Heidi

> Thanks for your help.
> With kind regards
>
> Andreia
> --
> --------------------------------------------
> Andreia J. Amaral
> Unidade de Imunologia Clínica
> Instituto de Medicina Molecular
> Universidade de Lisboa
> email: andreiaamaral at fm.ul.pt
>          andreia.fonseca at gmail.com
>