Your post makes absolutely no sense. A transaction, which is 4 characters by the way, is invoked from a terminal (or a program). That transaction is tied to a program, which is started when the transaction is invoked. Programs can in turn start a transaction (which invokes ANOTHER program), but a program does not ever start the process.
So assuming that you meant transaction T is invoking program A, how does program B get going -- LINK, XCTL, START, or some other method? And why on earth would you be using option 1.6 of the XPED transaction? Option 2 is Debugging Facilities and allows you to select the program that you want to debug. You can pull up the source display (assuming there is a DDIO file for the region that contains the current source code for program B), set breakpoints, and exit the XPED transaction to then start up your transaction T and get into debugging.