It is important to do the entire process inside a virtual machine to avoid any additional file infections and/or suffer the malicious effects of the virus.
Finding the original entry point
This is not a trivial task, since once the PE's entry point changes it is no longer recorded anywhere in the PR header. The only way to find the original entry point is through debugging the PE's execution throughout the malicious code added, until the additional code reaches to the original entry point.
Doing that depends heavily on how complex and protected the malicious code is. A simple approach (that might work) could be using a debugger such as ollydbg to run the code until it reached the original code's memory region (this can be done by opening the memory window and pressing f12 after selecting the relevant memory regions).
If the malicious code implements any anti-debugging protections you might need to bypass those.
Changing the entry point back
If you've used ollydbg in the previous part, you could use PE dump plugin (OllyDumpEx for example) to dump the PE with the entry point by using the plugin when EIP points to that entry point.
Alternatively, you can use any PE editing tool (like PEExplorer, LordPE or CFFExplorer) or advanced hex editor (My personal preference is 010Editor) to edit the Entry Point field in the PE header directly.
Although the malicious code should no longer run it is still inside your file. As an optional third step, you might want to remove the file infector's residue. According to the error message, it seems the file infector added it's own section, so by using a PE editor you could remove that new section in order to remove at least most of the virus's residues in your file.