Experimental and theoretical pIs do not match in most cases. Expasy or similar prediction tools give insights into the acidity of the target peptide or protein. Most of the adducts on the basic chain of the protein sequence that is known as PTMs can affect the total charge and can shift the readings/mobility in pI measurements. In silico tools might produce similar results, but they can be different in a real-world scenario. As befits the name, they are prediction algorithms. If you have very similar pIs for two different proteins, you should experimentally record their pIs w/o relying upon the outputs from Expasy.

Dear Wei Lin I highly recommend the following paper:

Article IPC – Isoelectric Point Calculator

Best regards.