I think SimShade made a beautiful recap but there is something that may be useful to add. I am an MD but not an immunologist so someone could sure be more accurate. It is true that there may be some variability in our self molecules/proteins/antigens (due to SNPs and other reasons). However the main parts (those with an active function such as enzymatic activity, receptor function etc) are usually conserved among individuals and sometimes among species. A lot of variability is found in the MHC molecules that have a key role in antigen recognition and presentation. This can explain in part why autoimmune disease develop more frequently in some individuals and the association with the HLA aplotype. Moreover, often when autoantibodies develop against some self-antigens (ex. anti-insulin antobodies or anti-TPO antibodies etc) there are different antibodies against different epitopes of the same molecule (polyclonal response). This is probably one of the main reasons why the tests for autoimmunity are quite accurate (because we usually don’t test for epitope specific antibodies even if it can be done using some specific tests). I hope this helps but a very useful test is: “Cellular and Molecular Immunology” by Abbas et al.