[BioC] How to cope with arrays hybridized at significantly different time.

Paparountas Triantafyllos t_paparountas at bioinformatics.gr
Mon Mar 16 12:55:16 CET 2009

Thank you for all the really nice information. 

I really believe that experimental design prior to the initiation of sample
harvesting is ideal as it minimizes the expenses.Currently we do use the block
technique whereupon we try to collect (at each round of sample collection)
samples from all different conditions that seem important. 

If we assume that we can harvest 3 viable samples of all 3 conditions that we
want to test in 1 week , and the target is to get 30 samples for each condition
it is easy to see that we need 30 weeks in order to get the total number of
arrays and thus the inter-batch variability will be high and the batch effects
will be very difficult to be removed. 

How do you propose to design or analyze such an experiment? Do you have any
similar experience ? 

The limitations are
1. Patients for the condition we try to test are few : 3 incidences/week
2. Cell isolation in blood is very time consuming which means that no more than
8 arrays can be processed every week 
3. We cannot store RNA  for many days so that we can hybridize arrays lets say
in 2 batches.

As I see this batch effect is inevitable. Frozen RMA might be a solution.

Thanks to everyone.

-Triantafyllos (Andy) Paparountas

More information about the Bioconductor mailing list