I guess you don't mean "probes" but the signal intensity values obtained from features ("spots") with such probes.

Answer: as usual. Average = arithmetic mean = sum divided by the number of summands.

But: the average is just not a good estimator because the distribution is skewed. For theoretical and mathematical reasons is is better to work with log fold-changes (LFC). You can simply average the LFCs. If you would antilog the mean LFC, the result would be the geometrical mean of the fold-changes.

There are often several features for the same gene. Some of these features may have the same probe sequences. Others might have differing sequences. You should think about what averages will be resonable (gene-wise with all different probe sequences, or probe-wise for all spots with the same probe sequence).

Thanks Jochen,

actually i am doing Array CGH analysis and i want to know the actual amplification and deltion of a particular region in chromosomes.for that probes, which we have used are tiling array probes, each probes are 56 bp apart which will cover whole chromosome, each probe has some intensity, so by simply doing log base2 change of intensity ratios will not suffice our need, i want to understand what will be the best method to get the actual amplification or how to set a threshold by which we can say that above threshold X its amplification, and below is deletion. can you suggest any method, which will suffice my need...