You can use any statistical tests made for this, depending on the availability of (biological or technical) replicates. The use of a threshold in differential gene expression analysis in principle has nothing to do with flagging a gene as differentially expressed. I mean, basic functioning of methods for deg identification is: use the gene expression levels in replicates to estimate a distribution of gene expression for each gene and for each condition. then the distributions (or averages in a more simple case) for the same gene in two different conditions are compared to see if they are significantly different. In case of very narrow distributions you can even identify as differentially expressed a gene with a small fold change, but in general a filter is applied to to the list of degs to select only those with a marked (with the assumption - false - that marked means biologically meaningful, in reality, we don't know) change. this is arbitrary, tough.

I ask to myself why you should expect smaller differences at the protein than at the transcript level. I would say, it depends.

Another approach would be to get the distribution of all changes, and if it is normal, use a Z-score transformation to get the proteins that changes the most. but this might give you a different answer, as the dispersion of the level of different proteins is different, irrespectively of their regulation.

Does any one have any idea about Perseus from MaxPlank, Can I use it here??