How can I compare proteomes that differ massively?
I am interested in comparing two fluids that I have proteomic data for two conditions (currently 3 replicates but I could increase this) in which the abundance of most proteins differs (i.e. that violates the assumptions of in silico normalization approaches (that I am aware of)). Is there a statistically robust way that I can compare relative abundance of the proteins between conditions?
Approaches I have considered:
- Using standard normalisation e.g. median centering and scaling
- Subtracting the (log) abundance of a common protein or set of common proteins?
- Ranking proteins and using non-parametric tests e.g. Wilcoxon-Mann-Whitney
Many thanks
Thanks - I will look into this. I'm not sure that the summed abundance is necessarily proportionate to the amount injected for the two samples, but I don't think that there is an appropriate way around this.
The summed abundance is not related to the absolute total amount if you express it as proportion of the total. The sum of all proteins will be 100%, irrespective of what amount you started with. In this way, you only get relative information. If preserving absolute information on abundance is necessary, don't go down this road. You mentioned relative abundance so my understanding was that you don't care about the absolute values.
Yes, that is true. I think my concern is less of a statistical/bioinformatic one and I'm sure that there is no way around it - given the gross differences is proteomes, some of the observed differences in abundance may be due to factors e.g. competitive ionisation that might influence the abundance of a protein in one condition but not the other.