triplex formation is strongly sequence dependent, so I would say, even if a triplex formation is occurring, it would be only partial, and the rest of the eRNA would have to be sticking out. But triplex formation is possible given that the circumstances are met. 

Osman: Since the entire length of the eRNA transcript will be complementary to the template, sequence requirement is not a concern to form a triplex. What might complicate the matter for triplex formation are (i)  the thermodynamic requirements to make very long stretches (>500bp) of triplexes inside the nucleus, as opposed to short ~20-30bp stretches that we encounter during gene silencing or in in vitro reactions, and (ii) the accessibility constraints imposed by the chromatin; and, perhaps to a lesser extent, (iii) myriads of transcription factors that gather around enhancers. I assume, therefore, that long stretches of DNA-eRNA triplexes would not form commonly.

Anil: I would have to disagree. not necessarily you can form a triplex just with an identical third strand. 1st- the duplex has to have at least 20bp long homopurine/homopyrimidine sequence. any inconsistency of the this rule wud not be favored. 2nd- triplexes may be destabilised if the Gs and As are too mixed because it creates the fluctuation in the backbone position of the third strand.

Osman, I guess we are digressing; but we both seem to agree that such long stretches of triplexes would be disfavored. Right? Now, let's move on to share if we know of any information on the other two possibilities: detachment or attachment. Again, if the eRNA remains attached at one end (arguably the 5' end) it might involve a short stretch of triplex--stabilized by proteins.

:) well, yes, a eRNA-long triplex formation is very unlikely but as you said; with the assistance of proteins triplexes can form at one end and be free floating at the other. I think triplex is more plausible than displacement of one DNA strand at one end to form a RNA-DNA duplex.