Well, the combination of sounds g+a+n does not sound particularly distinctive for me to be exclusively attributable to one sole language family. The same combinations of phonemes or very similar phonemes can be used by totally unrelated language families as affixes or other kinds of particles, generally to serve different semantic purposes, but random coincidences also happen here and there.
Besides, there is the one little big problem: are those surnames even like that, ending in -gan, in Irish itself? You're comparing Anglicized versions of those Irish surnames. I'm not totally sure of it, but after googling a bit the first sources seem to indicate those surnames end in different ways in Irish itself in their original forms: Mac an Mhadaidh (Maddigan), Ó Ríogáin, Ó Corragáin. All of them also have very plausible Celtic etymologies and relate to Celtic clans, so looking to a remote Turkic connection (one ancient enough for nearly all evidences of it to have disappeared) because of one little ending particle that looks similar to one found in Turkic is just an unnecessary, though "creative", exercise. Linguistics is a lot about being parsimonious: if you have a less extraordinary and more feasible explanation that explains the phenomenon, you generally don't need to look further to much more imaginative and unlikely hypotheses.