[BioC] cellHTS2 - 2 channel batch analysis

Wolfgang Huber huber at ebi.ac.uk
Wed Oct 29 16:15:59 CET 2008

Dear Steve

in cellHTS2, the basic data class is "cellHTS", which inherits from
"NChannelSet" (defined in Biobase).

You would consider your first round of the experiment as an NChannelSet
with N=2 (colours), 3*96=288 rows ("features") and 4 columns (2 for C
and 2 for E). You denote the metadata about the columns in the
"phenoData" slot. The metadata about the rows (well coordinate, plate
number, etc.) goes into the "featureData" slot.

There is no off-the-shelf way to use the same instance of an
"NChannelSet" for experiments in which people changed the plate
configuration mid-experiment. This is similar to that you cannot use a
single ExpressionSet instance for data from different types of
microarrays. For that, you will need to create two separate instances,
and deal with the combination, averaging, etc. by writing your own code.

Hope this helps.
Best wishes

Wolfgang Huber  EBI/EMBL  Cambridge UK  http://www.ebi.ac.uk/huber

24/09/2008 14:07 Steve Taylor scripsit
> Hi,
> I am trying to set up a two channel analysis using cellHTS2.
> I have cell line X. X has control (C) and experiment (E) conditions. The
> screen uses 3 x 96 well plates and C and E have 2 replicates each, hence
> C and E have 6 plates each in total.
> It gets complicated because there is a repeat of the whole experiment
> (on a separate day) and the experimenter used a different plate
> configuration. Also 3 of the control plates were not run, so for the
> second experimental run C has 3 plates and E has 6 plates.
> I'd like to combine both experiments but wondered what is the best way?
> I read you can normalize for batch effects which looks like it should
> help but on the whole I am unsure how to configure this analysis.
> Thanks for any help,
> Steve
> ------------------------------------------------------------------
> Medical Sciences Division
> Weatherall Institute of Molecular Medicine/Sir William Dunn School
> Oxford University

More information about the Bioconductor mailing list