[BioC] batch info for cellHTS2

[Gregoire Pau]

Dear Keren,

As mentioned in the cellHTS2 vignette, the variance adjustement is a 
final pre-processing step of the normalizePlates() function, consisting 
in dividing the measurements by a mesure of dispersion (sd, mad, iqr...) 
computed on groups of measurements (usually plates or batches of 
plates). The goal is to adjust the variance of the measurements between 
the different groups, which may differ due to different experimental 
conditions. Variance adjustment is a fine tuning step and you may not 
need it at this early analysis stage.

What do you mean by "combining" the results ? Depending on the 
biological question you want to answer, it could mean different things. 
  Is it a a differential expression analysis ? Is it a two-channel 
experiment, a transport assay ? CellHTS2 is able to analyze multichannel 
assays but the variates should be summarized at some point to univariate 
values with summarizeChannels().

Best regards,

Greg
---
Gregoire Pau
EMBL Research Officer
http://www.ebi.ac.uk/~gpau/


Baranes-Bachar, Keren (NIH/NCI) [V] wrote:
> Hello,
> I'm a NIH scientists that is doing a HTS experiment and I'm using your Cell HTS2 software to analyze our results.
> I'm doing two different experiments with 2 repeats for each of them. I want to combine the results and thought to try the batch function.
> Currently, I'm doing per plate normalization and no variance adjustment. I would like to do batch variance adjust, but I'm uncertain as to what this will do to the normalized results. Could you please provide me with a general explanation of variance adjustment, what it does and why?
> Additionally, when I add the batch number to the plate list and ask the program to do a variance adjust by batch, we get the following error:
> 
>> x <- configure(x,
> +                descripFile="Description.txt",
> +                confFile="Plateconf.txt",
> +                path=dataPath)
>> xn <- normalizePlates(x,
> +                       scale="multiplicative",
> +                       log=FALSE,
> +                       method="median",
> +                       varianceAdjust="byBatch")
> 
> Error in adjustVariancebyBatch(<S4 object of class "cellHTS">) :
>   'batch' should have dimensions 'Features x Samples x Channels' (768 x 4 x 1).
> In addition: Warning message:
> In dim(bb) != d :
>   longer object length is not a multiple of shorter object length
> 
> The error is not in the platelist file because the program identifies the different batches:
> 
>> batch(x)
>        replicate1 replicate2 replicate3 replicate4
> plate1          1          1          2          2
> plate2          1          1          2          2
> 
> Could you help me solve this error?
> 
> Also, please add any example scripts of a variance adjustment by batch and an explanation.
> Thank you in advance,
> Keren Baranes Bachar
> NCI/NIH
> 
> 	[[alternative HTML version deleted]]
> 
> _______________________________________________
> Bioconductor mailing list
> Bioconductor at stat.math.ethz.ch
> https://stat.ethz.ch/mailman/listinfo/bioconductor
> Search the archives: http://news.gmane.org/gmane.science.biology.informatics.conductor