Arkhangelsk,

A complete answer to your question(s) would require more time than I have to compose and perhaps more time than you would care to devote to reading, but in brief...  

The GKT, is a scientifically-based, statistically-sound paradigm (developed in one form by David Lykken in the late 1950s) that contains a key item and negative controls for each probed area and allows for a determination of information present or absent regarding some overall subject matter (e.g., overall examinee/suspect knowledge of the salient details of a crime scene) for a given examinee using autonomic measures.

A POT test is a presumably well-intentioned test for the presence of information developed by the polygraph community that fails a variety of statistical requirements for meaningful analysis.  In order for meaningful analysis the various POT items (or GKT items for that matter) have to be independent of one another and not ordered in a forward or reverse known sequence (violated within the POT format) and, of course, cannot be confounded with a lie test format, i.e., formulated in such a way as to require a yes or no answer which amounts to either a truthful or deceptive response (again violated with the POT format) to the posed question(s).

Although I believe the GKT exam using autonomic measures (contained within the standard polygraph instrument) to be sound, I believe a better way of probing for concealed or privileged information is to utilize the CNS, where memories are actually encoded and retrieved, and to use both positive and negative controls and using appropriate statistical analysis (e.g., bootstrapping) to determine whether the examinee’s responses to probed material is more similar to the positive or the negative controls.  

Most of my efforts of the last couple of decades have been devoted to this type of approach.

If memory serves correctly the MGQT in its original nomenclature referred to the Modified General Question Test.  It is a lie test that addresses multiple issues (frequently four) as opposed to another lie test format, e.g., the zone comparison test (ZCT) (which addresses only one area of interest); regardless of the specifics of the acronym, the MGQT is merely a lie test which addresses multiple issues poorly/inefficiently, whereas a ZCT is a format which addresses one issue poorly/inefficiently. Both and all other forms of lie detection are lacking in theoretical foundation and diagnostic validity.

Regards....