If I understand your question and graph correctly, your Y-axis is log(x/REF), where REF is some external standard. Your "Ref" on the x-axis you expect to be the same as REF, so that log(Ref/REF) "should be" zero, but you find it is not.
However, it looks like the mean(log(Ref/REF)) is still approximately zero. This is what I would expect: ...
What is Protein Expression Level?
This was the original title of the post, which I edited myself because I regard the answer as trivial, but the question as more substantial. To deal with the trivial first:
‘Level’ is not a scientific unit, and can only be used unambiguously
as a scientific term in its English sense in relation to liquids, e.g. “The level ...
If the query genome is unknown, a microarray cannot be made for a target species. Microarrays have DNA fragments of what you want to amplify on them. Those fragments must be known.
DNA microarrays are microscope slides that are printed with thousands
of tiny spots in defined positions, with each spot containing a known
DNA sequence or gene.
It is entirely possible that different cell-lines express the same genes at drastically different levels.
The proteinatlas provides data and analyses on differences between certain tissues or cell types / cell-lines.
If your cells are of the same line/tissue, then they might still differ dependent on the cell-cycle, age and external factors. There are ...
What you are looking at there is a microarray dataset (see the description here).
Microarrays are chips that have many sequence specific probes on them and sometimes they have multiple probes for the same gene.
Readout differences between the probes of the same gene could be caused by either noise in the data or biological effects like expression differences ...